| { |
| "metadata": { |
| "ParamSize": 709, |
| "ParamBytes": 18431289344.0, |
| "BitsPerParam": 3.917390901829854 |
| }, |
| "records": [ |
| { |
| "dataPath": "params_shard_0.bin", |
| "format": "raw-shard", |
| "nbytes": 389283840, |
| "records": [ |
| { |
| "name": "lm_head.q_weight", |
| "shape": [ |
| 152064, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 389283840, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "721ad6b72844b8957e2e85e8c37d10e8" |
| }, |
| { |
| "dataPath": "params_shard_1.bin", |
| "format": "raw-shard", |
| "nbytes": 48660480, |
| "records": [ |
| { |
| "name": "lm_head.q_scale", |
| "shape": [ |
| 152064, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 48660480, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "9eecc91b051414bfd055d9b5e2f7d933" |
| }, |
| { |
| "dataPath": "params_shard_2.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.58.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "78717bbba57eb3eba895630af32b93ee" |
| }, |
| { |
| "dataPath": "params_shard_3.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.58.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "857162b821362d7a39dcedaaa4d7b839" |
| }, |
| { |
| "dataPath": "params_shard_4.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.59.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "09383110a33fd1d0f54ffbaac372714c" |
| }, |
| { |
| "dataPath": "params_shard_5.bin", |
| "format": "raw-shard", |
| "nbytes": 26572800, |
| "records": [ |
| { |
| "name": "model.layers.58.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.58.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 10240 |
| }, |
| { |
| "name": "model.layers.58.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 8857600 |
| }, |
| { |
| "name": "model.layers.58.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26552320 |
| }, |
| { |
| "name": "model.layers.59.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26562560 |
| } |
| ], |
| "md5sum": "50f0a3e67c29527018aeb9dcd53fcd0b" |
| }, |
| { |
| "dataPath": "params_shard_6.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.59.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "2320c1db82279b73c92647fdfd85827e" |
| }, |
| { |
| "dataPath": "params_shard_7.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.59.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "aedd4c827930fc456328be3356f8f721" |
| }, |
| { |
| "dataPath": "params_shard_8.bin", |
| "format": "raw-shard", |
| "nbytes": 28860416, |
| "records": [ |
| { |
| "name": "model.layers.59.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.59.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 8847360 |
| }, |
| { |
| "name": "model.layers.59.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26542080 |
| }, |
| { |
| "name": "model.layers.59.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 26552320 |
| }, |
| { |
| "name": "model.layers.59.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 26566656 |
| } |
| ], |
| "md5sum": "b867ecc00d8e967acaa0d9fcb5092e2f" |
| }, |
| { |
| "dataPath": "params_shard_9.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.60.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "40425654f1d7fda3ff1f360c44334571" |
| }, |
| { |
| "dataPath": "params_shard_10.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.60.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8b321fcee8f6a0ea9b4fc3ffac0502aa" |
| }, |
| { |
| "dataPath": "params_shard_11.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.60.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8f220fb4c6201bb64fcc2bf4a724a0ae" |
| }, |
| { |
| "dataPath": "params_shard_12.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.60.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "89e6959bf8178bbb0301bba096f248f3" |
| }, |
| { |
| "dataPath": "params_shard_13.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.59.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.59.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.60.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.60.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.60.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.60.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.60.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "b7d1cc98093a4a14b297e729d9ab7ca7" |
| }, |
| { |
| "dataPath": "params_shard_14.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.61.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d7b132a88617e32fcda4f2f808d0c32c" |
| }, |
| { |
| "dataPath": "params_shard_15.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.61.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8ad429cb7b62fc365d191c6a86f40bb3" |
| }, |
| { |
| "dataPath": "params_shard_16.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.61.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "27e03c9cac1f64ead460ce48c9f9a03b" |
| }, |
| { |
| "dataPath": "params_shard_17.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.61.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "88fcb9ac63e19bb0befa8e2d40e2fb05" |
| }, |
| { |
| "dataPath": "params_shard_18.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.60.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.60.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.61.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.61.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.61.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.61.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.61.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "224c8e92910cffe3ecaec0f9dcc33c04" |
| }, |
| { |
| "dataPath": "params_shard_19.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.62.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "69a05c59a0d5c3eb17fc705617b1d529" |
| }, |
| { |
| "dataPath": "params_shard_20.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.62.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "726820b1cb7386ead67b6d8660762b81" |
| }, |
| { |
| "dataPath": "params_shard_21.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.62.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "01db479d281beac65a09e151eaaed772" |
| }, |
| { |
| "dataPath": "params_shard_22.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.62.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "36bdb477e2a0b1ff4988f3afb10a22f6" |
| }, |
| { |
| "dataPath": "params_shard_23.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.61.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.61.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.62.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.62.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.62.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.62.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.62.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "c15c18b8a0717a87a6f34c5c316d200e" |
| }, |
| { |
| "dataPath": "params_shard_24.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.63.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "94fd5fe954dab56c82b620ab1c00e7b8" |
| }, |
| { |
| "dataPath": "params_shard_25.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.63.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "28ce6237e78da5ca539bb6f0e29a14bc" |
| }, |
| { |
| "dataPath": "params_shard_26.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.63.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "64df29f479f37873c7e6ecee6af5ba62" |
| }, |
| { |
| "dataPath": "params_shard_27.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.63.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "45f61f903c3080921f436e1dc1c83a00" |
| }, |
| { |
| "dataPath": "params_shard_28.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.62.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.62.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.63.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.63.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.63.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.63.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.63.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "aafc94f947fd00c918adbf4c65829139" |
| }, |
| { |
| "dataPath": "params_shard_29.bin", |
| "format": "raw-shard", |
| "nbytes": 389283840, |
| "records": [ |
| { |
| "name": "model.embed_tokens.q_weight", |
| "shape": [ |
| 152064, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 389283840, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "6608733679830196474af4454d05e97a" |
| }, |
| { |
| "dataPath": "params_shard_30.bin", |
| "format": "raw-shard", |
| "nbytes": 48660480, |
| "records": [ |
| { |
| "name": "model.embed_tokens.q_scale", |
| "shape": [ |
| 152064, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 48660480, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "3ef669ef366e1032f52db5fb4f63ea71" |
| }, |
| { |
| "dataPath": "params_shard_31.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.0.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e14e0f0ee6b3c2b7db1e7c978a5e3191" |
| }, |
| { |
| "dataPath": "params_shard_32.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.0.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c9f3b88b82291cf0a66ee1d9228aec16" |
| }, |
| { |
| "dataPath": "params_shard_33.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.0.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5d2a587ad3461a136bd740ba0bf1ee54" |
| }, |
| { |
| "dataPath": "params_shard_34.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.0.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b9ace1284beb32ed2118cc9a3710af61" |
| }, |
| { |
| "dataPath": "params_shard_35.bin", |
| "format": "raw-shard", |
| "nbytes": 25931776, |
| "records": [ |
| { |
| "name": "model.layers.63.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.63.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.norm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.0.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.0.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14766080 |
| }, |
| { |
| "name": "model.layers.0.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.0.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23623680 |
| }, |
| { |
| "name": "model.layers.0.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23638016 |
| } |
| ], |
| "md5sum": "026c2176b10f5b30df135e7438b3b55d" |
| }, |
| { |
| "dataPath": "params_shard_36.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.1.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0cfeb4193aa6d220457eac4c75b15748" |
| }, |
| { |
| "dataPath": "params_shard_37.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.1.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ea0a07d6ae50ca2854673a8696e40fe6" |
| }, |
| { |
| "dataPath": "params_shard_38.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.1.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a5efeb149d6476de12ec6c44a374484c" |
| }, |
| { |
| "dataPath": "params_shard_39.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.1.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e1aee4bd2793572425b6c8185a87ca6a" |
| }, |
| { |
| "dataPath": "params_shard_40.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.0.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.0.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.1.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.1.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.1.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.1.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.1.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "280ac268543a024b58494114227a9561" |
| }, |
| { |
| "dataPath": "params_shard_41.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.2.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "6bb94fb9071f4b0c22dbb3ed69eee870" |
| }, |
| { |
| "dataPath": "params_shard_42.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.2.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "4c73d68a235fb80312a8db8f64a92c9f" |
| }, |
| { |
| "dataPath": "params_shard_43.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.2.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "72e7e67d08d83c0e5aaab5c4bafb981f" |
| }, |
| { |
| "dataPath": "params_shard_44.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.2.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b5d6eb10137bbe0f927d654a143e1de7" |
| }, |
| { |
| "dataPath": "params_shard_45.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.1.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.1.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.2.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.2.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.2.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.2.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.2.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "c74b142308db86979184e8f09fb7ee78" |
| }, |
| { |
| "dataPath": "params_shard_46.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.3.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "9f6bbe7fb469dd15d45735a1830429fc" |
| }, |
| { |
| "dataPath": "params_shard_47.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.3.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "bc7f4291de6c8b32846410cfb777c048" |
| }, |
| { |
| "dataPath": "params_shard_48.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.3.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5117873198bcf4128f32fba484653ed0" |
| }, |
| { |
| "dataPath": "params_shard_49.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.3.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "db358db9bde4d8f1fc0b52d428261298" |
| }, |
| { |
| "dataPath": "params_shard_50.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.2.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.2.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.3.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.3.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.3.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.3.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.3.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "e6ae6b8995488ce4a2ccbf7fb6b6ec0a" |
| }, |
| { |
| "dataPath": "params_shard_51.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.4.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5e1cb34f5d0db6ab19f52bbfaa636e71" |
| }, |
| { |
| "dataPath": "params_shard_52.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.4.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "59ac55f05a040bdc1242f71049cdc948" |
| }, |
| { |
| "dataPath": "params_shard_53.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.4.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5326239b7cce6005e734302f5a388543" |
| }, |
| { |
| "dataPath": "params_shard_54.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.4.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e6866e89882d6162c697c99c46c83240" |
| }, |
| { |
| "dataPath": "params_shard_55.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.3.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.3.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.4.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.4.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.4.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.4.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.4.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "69dae9396f45286efc47fedbacbcd12c" |
| }, |
| { |
| "dataPath": "params_shard_56.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.5.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b90237db72aaf0ec78e5c7ef2aac7c93" |
| }, |
| { |
| "dataPath": "params_shard_57.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.5.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c08abec760a0770f3d46fa5db00d8369" |
| }, |
| { |
| "dataPath": "params_shard_58.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.5.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "4b47151e4249e743c6bf90348c8c65fa" |
| }, |
| { |
| "dataPath": "params_shard_59.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.5.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0f4c2b30592e99df87f09440f31c4fb3" |
| }, |
| { |
| "dataPath": "params_shard_60.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.4.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.4.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.5.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.5.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.5.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.5.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.5.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "093ba478363a09977cdc04bf5720e1de" |
| }, |
| { |
| "dataPath": "params_shard_61.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.6.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "df3944689a83b3c7682b0870d1c49054" |
| }, |
| { |
| "dataPath": "params_shard_62.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.6.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "225c30cb21cfff15c446e741787ec2b9" |
| }, |
| { |
| "dataPath": "params_shard_63.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.6.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "59cd1010a06f95497d58e60130cbc982" |
| }, |
| { |
| "dataPath": "params_shard_64.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.6.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5bbe697eac8184931fb04cb2e4830205" |
| }, |
| { |
| "dataPath": "params_shard_65.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.5.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.5.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.6.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.6.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.6.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.6.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.6.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "a5aaa004010041bd7d13175688c49b18" |
| }, |
| { |
| "dataPath": "params_shard_66.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.7.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0f3f730e82d53e4d5e7b2ab2eecf0ea5" |
| }, |
| { |
| "dataPath": "params_shard_67.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.7.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5b0eb39efb7b275bd06f9248a4e6cc56" |
| }, |
| { |
| "dataPath": "params_shard_68.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.7.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ec360a226468c5c100aa9759b2f173c3" |
| }, |
| { |
| "dataPath": "params_shard_69.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.7.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5f725b4a482e91080c915a53ca7188f1" |
| }, |
| { |
| "dataPath": "params_shard_70.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.6.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.6.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.7.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.7.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.7.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.7.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.7.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "4462174195214f1fde83ae587e72395c" |
| }, |
| { |
| "dataPath": "params_shard_71.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.8.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a12a2c8c07dd2d57335c07150c7b5912" |
| }, |
| { |
| "dataPath": "params_shard_72.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.8.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "9c59d2cdf4c80d19059795f607faee69" |
| }, |
| { |
| "dataPath": "params_shard_73.bin", |
| "format": "raw-shard", |
| "nbytes": 32454656, |
| "records": [ |
| { |
| "name": "model.layers.7.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.7.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.8.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.8.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 32440320 |
| } |
| ], |
| "md5sum": "043df241dd6bcb2459b001957a75d07d" |
| }, |
| { |
| "dataPath": "params_shard_74.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.10.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1354b5bb3e0d3888feebca532a0891cb" |
| }, |
| { |
| "dataPath": "params_shard_75.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.10.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "4e4d2fa9250f724d030867badce69073" |
| }, |
| { |
| "dataPath": "params_shard_76.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.10.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "109c2017b6e375f0bfcbedaf4325d43d" |
| }, |
| { |
| "dataPath": "params_shard_77.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.10.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f6a9071bdeb76dfc803956384e5fd243" |
| }, |
| { |
| "dataPath": "params_shard_78.bin", |
| "format": "raw-shard", |
| "nbytes": 28215296, |
| "records": [ |
| { |
| "name": "model.layers.8.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.8.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.8.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| }, |
| { |
| "name": "model.layers.10.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 17039360 |
| }, |
| { |
| "name": "model.layers.10.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 17049600 |
| }, |
| { |
| "name": "model.layers.10.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25896960 |
| }, |
| { |
| "name": "model.layers.10.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 25907200 |
| }, |
| { |
| "name": "model.layers.10.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 25921536 |
| } |
| ], |
| "md5sum": "1fe9ce270648414d523be7d86799f284" |
| }, |
| { |
| "dataPath": "params_shard_79.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.11.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "547bd41037ad9f0dc09a18167e4ed5b0" |
| }, |
| { |
| "dataPath": "params_shard_80.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.11.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e6235c3f1e1baadb7c442c2320468a4a" |
| }, |
| { |
| "dataPath": "params_shard_81.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.11.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "055b1d1599ad782e4c262174c481700c" |
| }, |
| { |
| "dataPath": "params_shard_82.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.11.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b75b72e5d4d0b35eff67e9a18ee40e4f" |
| }, |
| { |
| "dataPath": "params_shard_83.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.10.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.10.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.11.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.11.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.11.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.11.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.11.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "0fa1de2d55b713895ad77e71dc556288" |
| }, |
| { |
| "dataPath": "params_shard_84.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.12.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f8d5c5a752b1a27d53f5e6b6fdec4f50" |
| }, |
| { |
| "dataPath": "params_shard_85.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.12.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "cd74be04cb5758e0a2bbf85e88b19430" |
| }, |
| { |
| "dataPath": "params_shard_86.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.12.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "850a52f2c79fbfe722506a554f8c29d0" |
| }, |
| { |
| "dataPath": "params_shard_87.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.12.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1e0e53fafb8260f5f3b9a8bb78207b5e" |
| }, |
| { |
| "dataPath": "params_shard_88.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.11.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.11.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.12.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.12.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.12.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.12.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.12.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "b8e4e2744a82dff23e9b3aeee38c4a51" |
| }, |
| { |
| "dataPath": "params_shard_89.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.13.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "22e07cf8b330ef96a6186c053b2e207c" |
| }, |
| { |
| "dataPath": "params_shard_90.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.13.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c468aca8595542a1fb802838e105d309" |
| }, |
| { |
| "dataPath": "params_shard_91.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.13.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "06f65fc92830917315837e2a3fb351ed" |
| }, |
| { |
| "dataPath": "params_shard_92.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.13.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "96eb6faa828877bec3d6920a13b15009" |
| }, |
| { |
| "dataPath": "params_shard_93.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.12.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.12.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.13.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.13.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.13.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.13.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.13.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "43d827a328b651e3000224284a93199a" |
| }, |
| { |
| "dataPath": "params_shard_94.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.14.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f08e06cd5995fe85430adae73bbc619f" |
| }, |
| { |
| "dataPath": "params_shard_95.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.14.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e62ebe5496f30efda7d3abf67d901e27" |
| }, |
| { |
| "dataPath": "params_shard_96.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.14.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8cf9a8b2539ec8c09f0eb257ca6ae49b" |
| }, |
| { |
| "dataPath": "params_shard_97.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.14.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "3245dba5e8eab1ecf766038089d5c2c5" |
| }, |
| { |
| "dataPath": "params_shard_98.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.13.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.13.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.14.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.14.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.14.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.14.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.14.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "49057b3ec2d621abc6d0c74e7c375b91" |
| }, |
| { |
| "dataPath": "params_shard_99.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.15.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "76499526effdbafbd687ebf374532873" |
| }, |
| { |
| "dataPath": "params_shard_100.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.15.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5f3cea7118a557fe00fc70b17cd66ee8" |
| }, |
| { |
| "dataPath": "params_shard_101.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.15.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8a05bae50344504f6ac539351b60611e" |
| }, |
| { |
| "dataPath": "params_shard_102.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.15.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "08214974ebc1a61d4f5452bc4edfab65" |
| }, |
| { |
| "dataPath": "params_shard_103.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.14.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.14.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.15.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.15.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.15.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.15.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.15.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "ffa351b5ad1afc0525c24276ca8aeccb" |
| }, |
| { |
| "dataPath": "params_shard_104.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.16.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "edb758999ef160bee0a29f6075d943c5" |
| }, |
| { |
| "dataPath": "params_shard_105.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.16.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5a24f300837301a3742d9754ff427bc9" |
| }, |
| { |
| "dataPath": "params_shard_106.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.16.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "4c8c0baf28c83110a478aef3df3b7378" |
| }, |
| { |
| "dataPath": "params_shard_107.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.16.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "4019db6f479dd506d45b31fca2382174" |
| }, |
| { |
| "dataPath": "params_shard_108.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.15.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.15.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.16.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.16.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.16.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.16.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.16.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "69b06a8721338019fdb72d0e58f2e29b" |
| }, |
| { |
| "dataPath": "params_shard_109.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.17.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d53c1971b2efa6ba46fa0b91ca365da1" |
| }, |
| { |
| "dataPath": "params_shard_110.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.17.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8e4017e0be806e6905f5be63224a7e90" |
| }, |
| { |
| "dataPath": "params_shard_111.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.17.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8f3eb4ca7b3f0c9ff7dc181c18dde65d" |
| }, |
| { |
| "dataPath": "params_shard_112.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.17.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "467b2f58e7f454371385a242b27a5785" |
| }, |
| { |
| "dataPath": "params_shard_113.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.16.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.16.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.17.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.17.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.17.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.17.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.17.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "b2629bd9a5854b88f505b25defa37273" |
| }, |
| { |
| "dataPath": "params_shard_114.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.18.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "3ed21c31bd280a4d3df7bcbbd6d160e6" |
| }, |
| { |
| "dataPath": "params_shard_115.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.18.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d1feb3e49d21487c7c358c38bd12efcb" |
| }, |
| { |
| "dataPath": "params_shard_116.bin", |
| "format": "raw-shard", |
| "nbytes": 32454656, |
| "records": [ |
| { |
| "name": "model.layers.17.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.17.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.18.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.18.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 32440320 |
| } |
| ], |
| "md5sum": "c3064faccf72ccc47830c1658ad3e126" |
| }, |
| { |
| "dataPath": "params_shard_117.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.8.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d785581e8f63999331254c13ea1ff41b" |
| }, |
| { |
| "dataPath": "params_shard_118.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.9.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "74ab16c8a23dc64c348ce345a78d7bc1" |
| }, |
| { |
| "dataPath": "params_shard_119.bin", |
| "format": "raw-shard", |
| "nbytes": 25917440, |
| "records": [ |
| { |
| "name": "model.layers.18.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.18.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.18.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| }, |
| { |
| "name": "model.layers.8.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 17039360 |
| }, |
| { |
| "name": "model.layers.8.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 17049600 |
| }, |
| { |
| "name": "model.layers.8.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25896960 |
| }, |
| { |
| "name": "model.layers.9.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25907200 |
| } |
| ], |
| "md5sum": "0c4762a9a5517ef971a14aaf87719151" |
| }, |
| { |
| "dataPath": "params_shard_120.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.9.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d56aec896959f2c941a13f1babbd1eca" |
| }, |
| { |
| "dataPath": "params_shard_121.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.9.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "eaee514808d8e268d2cffa391c2b5d91" |
| }, |
| { |
| "dataPath": "params_shard_122.bin", |
| "format": "raw-shard", |
| "nbytes": 28860416, |
| "records": [ |
| { |
| "name": "model.layers.9.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.9.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 8847360 |
| }, |
| { |
| "name": "model.layers.9.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26542080 |
| }, |
| { |
| "name": "model.layers.9.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 26552320 |
| }, |
| { |
| "name": "model.layers.9.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 26566656 |
| } |
| ], |
| "md5sum": "3ba487ca093f06121c20b338c8fca85e" |
| }, |
| { |
| "dataPath": "params_shard_123.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.18.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c6eb3ce74a92230d34d410092f45772b" |
| }, |
| { |
| "dataPath": "params_shard_124.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.19.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "699cd891e64a547921f7cd2f7af6957e" |
| }, |
| { |
| "dataPath": "params_shard_125.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.19.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f63ed904b839084ee0be9f4faf3eedfd" |
| }, |
| { |
| "dataPath": "params_shard_126.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.19.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "25abd84798facb4f419bfd6c6eb24b1f" |
| }, |
| { |
| "dataPath": "params_shard_127.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.19.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "9557e21a32a4c148dd61e83ae10e4acc" |
| }, |
| { |
| "dataPath": "params_shard_128.bin", |
| "format": "raw-shard", |
| "nbytes": 32495616, |
| "records": [ |
| { |
| "name": "model.layers.9.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.9.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.18.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.18.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.18.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.19.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.19.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 23623680 |
| }, |
| { |
| "name": "model.layers.19.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 32471040 |
| }, |
| { |
| "name": "model.layers.19.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 32481280 |
| } |
| ], |
| "md5sum": "771f58035e34038ab2994b99051c0756" |
| }, |
| { |
| "dataPath": "params_shard_129.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.20.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0acbe86f7c60b669ee31330ad1b731dc" |
| }, |
| { |
| "dataPath": "params_shard_130.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.20.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1ad64dadc04de20a5bc3c8943f87492a" |
| }, |
| { |
| "dataPath": "params_shard_131.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.20.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "2d36c4bbcf42ecc5f164621c588d6905" |
| }, |
| { |
| "dataPath": "params_shard_132.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.20.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0cf9ac79edfd2b03299dbf2bddd4662e" |
| }, |
| { |
| "dataPath": "params_shard_133.bin", |
| "format": "raw-shard", |
| "nbytes": 28215296, |
| "records": [ |
| { |
| "name": "model.layers.19.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.19.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.19.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| }, |
| { |
| "name": "model.layers.20.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 17039360 |
| }, |
| { |
| "name": "model.layers.20.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 17049600 |
| }, |
| { |
| "name": "model.layers.20.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25896960 |
| }, |
| { |
| "name": "model.layers.20.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 25907200 |
| }, |
| { |
| "name": "model.layers.20.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 25921536 |
| } |
| ], |
| "md5sum": "018d0845d34bf6289b40594615a3bba8" |
| }, |
| { |
| "dataPath": "params_shard_134.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.21.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "32ce40ad7acaa9924b545dec8bbc8805" |
| }, |
| { |
| "dataPath": "params_shard_135.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.21.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "64d644e13353fd86b3cf0c00a39bab63" |
| }, |
| { |
| "dataPath": "params_shard_136.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.21.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "50b79f1fc3a5f0242e2a3c62d8815bf4" |
| }, |
| { |
| "dataPath": "params_shard_137.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.21.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b95c8b3bdfce60778d7e95fe7410a243" |
| }, |
| { |
| "dataPath": "params_shard_138.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.20.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.20.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.21.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.21.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.21.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.21.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.21.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "f9cc29e938d28dc05ebf7ec408d7eb19" |
| }, |
| { |
| "dataPath": "params_shard_139.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.22.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "389cc0fa1f936c8b3d3dd2ad37860b9d" |
| }, |
| { |
| "dataPath": "params_shard_140.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.22.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "55350b390c7f44c2453b1eda8492006f" |
| }, |
| { |
| "dataPath": "params_shard_141.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.22.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "287da3dd22fba05e920f4fbee9648723" |
| }, |
| { |
| "dataPath": "params_shard_142.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.22.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "209f77bd56053ebaf7962ef0c2adc1de" |
| }, |
| { |
| "dataPath": "params_shard_143.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.21.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.21.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.22.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.22.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.22.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.22.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.22.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "4ce1c219b34287b2be09cff946fb9d3d" |
| }, |
| { |
| "dataPath": "params_shard_144.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.23.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a38e53d30d78124168d0cf1992299a17" |
| }, |
| { |
| "dataPath": "params_shard_145.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.23.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b7c6666a205260c2b807887ca050bb49" |
| }, |
| { |
| "dataPath": "params_shard_146.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.23.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "289770e22db8c8741067618503bdfe92" |
| }, |
| { |
| "dataPath": "params_shard_147.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.23.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8a2bcb0e1c8051ba4803e58752d4a045" |
| }, |
| { |
| "dataPath": "params_shard_148.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.22.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.22.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.23.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.23.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.23.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.23.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.23.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "5206cdb4946c9d9e5b4a82fb3ec2d3e7" |
| }, |
| { |
| "dataPath": "params_shard_149.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.24.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "95ad0afb5197f67e0ee504425f04cc77" |
| }, |
| { |
| "dataPath": "params_shard_150.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.24.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "db8db5b4ab60fa526b12769983bf902b" |
| }, |
| { |
| "dataPath": "params_shard_151.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.24.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "bf05d0de2b052f61e5ebc5c97f418e2a" |
| }, |
| { |
| "dataPath": "params_shard_152.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.24.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a3a9ea0d2e6cd1ca5d979cd4ddf7b9cc" |
| }, |
| { |
| "dataPath": "params_shard_153.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.23.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.23.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.24.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.24.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.24.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.24.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.24.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "5ad32b2d8bac599746aff2d399b29715" |
| }, |
| { |
| "dataPath": "params_shard_154.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.25.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f32004e6aa10ea42f213ce0e4c26154a" |
| }, |
| { |
| "dataPath": "params_shard_155.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.25.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "03f5a32935110abe1be27920b8d3cfc9" |
| }, |
| { |
| "dataPath": "params_shard_156.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.25.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "18419da4ecbbdece8dee41ce4ba59672" |
| }, |
| { |
| "dataPath": "params_shard_157.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.25.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "123c3321fa1daac983f2805b92cf638e" |
| }, |
| { |
| "dataPath": "params_shard_158.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.24.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.24.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.25.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.25.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.25.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.25.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.25.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "cca289e9bbf144a0823eb6833c87de00" |
| }, |
| { |
| "dataPath": "params_shard_159.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.26.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b56da4a86a2de667abc32baebd8d6464" |
| }, |
| { |
| "dataPath": "params_shard_160.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.26.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "17636106642e9d2f5db6c63dca61c130" |
| }, |
| { |
| "dataPath": "params_shard_161.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.26.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "222ba201879e1eae98fc818b2c1d978c" |
| }, |
| { |
| "dataPath": "params_shard_162.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.26.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "94803ff7e857a42959b8c93eb9010b04" |
| }, |
| { |
| "dataPath": "params_shard_163.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.25.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.25.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.26.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.26.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.26.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.26.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.26.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "900ab5c2f0773b3b6ce11e1ddaeec19d" |
| }, |
| { |
| "dataPath": "params_shard_164.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.27.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "45eda3a12dddf81deedd60cbb632afa5" |
| }, |
| { |
| "dataPath": "params_shard_165.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.27.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f441012d659738da049f332cb66b5e3e" |
| }, |
| { |
| "dataPath": "params_shard_166.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.27.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "eddfa89c651b3a63ebf9450984fb3ea4" |
| }, |
| { |
| "dataPath": "params_shard_167.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.27.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c38c49e90bd8622ee9bbd44abf858aa1" |
| }, |
| { |
| "dataPath": "params_shard_168.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.26.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.26.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.27.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.27.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.27.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.27.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.27.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "e7e1d3376af8e64dac9847d65141b861" |
| }, |
| { |
| "dataPath": "params_shard_169.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.28.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ea3cbb60ccd62da129ab1c0c30ad0bc8" |
| }, |
| { |
| "dataPath": "params_shard_170.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.28.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d1f28623a8495c44f9f7679e2e737932" |
| }, |
| { |
| "dataPath": "params_shard_171.bin", |
| "format": "raw-shard", |
| "nbytes": 32454656, |
| "records": [ |
| { |
| "name": "model.layers.27.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.27.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.28.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.28.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 32440320 |
| } |
| ], |
| "md5sum": "1a85a93beb6c90cb7b5d80833f9597e9" |
| }, |
| { |
| "dataPath": "params_shard_172.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.28.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "6d386474ac8cb83b50c376759b8c6d3c" |
| }, |
| { |
| "dataPath": "params_shard_173.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.29.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "3955892c6a4fbb0c42a0dc8ed2fa9d1a" |
| }, |
| { |
| "dataPath": "params_shard_174.bin", |
| "format": "raw-shard", |
| "nbytes": 25917440, |
| "records": [ |
| { |
| "name": "model.layers.28.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.28.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.28.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| }, |
| { |
| "name": "model.layers.28.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 17039360 |
| }, |
| { |
| "name": "model.layers.28.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 17049600 |
| }, |
| { |
| "name": "model.layers.28.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25896960 |
| }, |
| { |
| "name": "model.layers.29.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25907200 |
| } |
| ], |
| "md5sum": "7032997813333562b6dd322e10e0481e" |
| }, |
| { |
| "dataPath": "params_shard_175.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.29.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "334aa6367cd48007fc7654932223bad6" |
| }, |
| { |
| "dataPath": "params_shard_176.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.29.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "15815d047b296fb424e0ab2d5df5395f" |
| }, |
| { |
| "dataPath": "params_shard_177.bin", |
| "format": "raw-shard", |
| "nbytes": 28860416, |
| "records": [ |
| { |
| "name": "model.layers.29.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.29.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 8847360 |
| }, |
| { |
| "name": "model.layers.29.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26542080 |
| }, |
| { |
| "name": "model.layers.29.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 26552320 |
| }, |
| { |
| "name": "model.layers.29.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 26566656 |
| } |
| ], |
| "md5sum": "dc1fdad0fba6e46a67de2fedd4fbc234" |
| }, |
| { |
| "dataPath": "params_shard_178.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.30.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e02e54e600e99e2affe326fc593b880a" |
| }, |
| { |
| "dataPath": "params_shard_179.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.30.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "34100218b381f206b9c41277cf180e71" |
| }, |
| { |
| "dataPath": "params_shard_180.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.30.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f5feb9677fc20608b9e5636aa4629227" |
| }, |
| { |
| "dataPath": "params_shard_181.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.30.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "70e990834ba1bcaf0817b3764a9454fc" |
| }, |
| { |
| "dataPath": "params_shard_182.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.29.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.29.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.30.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.30.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.30.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.30.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.30.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "2106c0621abdc9051c24cec7dbeb63cf" |
| }, |
| { |
| "dataPath": "params_shard_183.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.31.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d4f7caa2af1a3b7bc1c232e2134c3e1e" |
| }, |
| { |
| "dataPath": "params_shard_184.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.31.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "aac7f1eb37bf627673fcaf53baa0e133" |
| }, |
| { |
| "dataPath": "params_shard_185.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.31.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "6d25a92b035f1fac86d58fd18356d3dc" |
| }, |
| { |
| "dataPath": "params_shard_186.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.31.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "2a1b46ee9fdf395126eef2419a6d1a56" |
| }, |
| { |
| "dataPath": "params_shard_187.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.30.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.30.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.31.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.31.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.31.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.31.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.31.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "96d74ac46c09a819b65b9df997e46af0" |
| }, |
| { |
| "dataPath": "params_shard_188.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.32.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "daf53d165634fdc41c1735cb682804c7" |
| }, |
| { |
| "dataPath": "params_shard_189.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.32.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8cd65b4468b41282adf20e8bf7b30719" |
| }, |
| { |
| "dataPath": "params_shard_190.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.32.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c26a468d4cf2edad034c9863bfafb6b1" |
| }, |
| { |
| "dataPath": "params_shard_191.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.32.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0a3f6e11319ca9841dd6ae37f12f1c92" |
| }, |
| { |
| "dataPath": "params_shard_192.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.31.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.31.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.32.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.32.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.32.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.32.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.32.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "d250f90e61a840b3b3f018d0e15f0f16" |
| }, |
| { |
| "dataPath": "params_shard_193.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.33.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "9fdc482ba7718f9d8d64afd44e2aa4ce" |
| }, |
| { |
| "dataPath": "params_shard_194.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.33.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "46b7e9c05912ccd33eaf92a5c1d28eb3" |
| }, |
| { |
| "dataPath": "params_shard_195.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.33.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "757ebfb137e16621314dbc88fc34ef20" |
| }, |
| { |
| "dataPath": "params_shard_196.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.33.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0bbc52e5d05eae716c2a41707b350bd0" |
| }, |
| { |
| "dataPath": "params_shard_197.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.32.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.32.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.33.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.33.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.33.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.33.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.33.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "71050ff0a4878ae24cdaa8a1ab212a47" |
| }, |
| { |
| "dataPath": "params_shard_198.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.34.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ddad69a1460a9b6a278ed54a42b8ffeb" |
| }, |
| { |
| "dataPath": "params_shard_199.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.34.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "2ddab42ecf3405de04be62543df65b85" |
| }, |
| { |
| "dataPath": "params_shard_200.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.34.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "98c440cd3dcb02fe65e5d4df815258cc" |
| }, |
| { |
| "dataPath": "params_shard_201.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.34.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "126d64d91fa623a107d12b8bea45fbc5" |
| }, |
| { |
| "dataPath": "params_shard_202.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.33.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.33.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.34.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.34.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.34.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.34.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.34.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "67c6532f188978ec66782c8a216c41dd" |
| }, |
| { |
| "dataPath": "params_shard_203.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.35.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1425e2d77b474ef3aacd6416390c199b" |
| }, |
| { |
| "dataPath": "params_shard_204.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.35.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f1362467e486773653014a78e2a28c9e" |
| }, |
| { |
| "dataPath": "params_shard_205.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.35.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "607c85582501a1ff1aa4bd8db8c08820" |
| }, |
| { |
| "dataPath": "params_shard_206.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.35.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "6c37dcbc6dfd21b6c469b24ba0941b55" |
| }, |
| { |
| "dataPath": "params_shard_207.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.34.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.34.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.35.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.35.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.35.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.35.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.35.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "22a36b8fd0463c1f6242e15c495a7ca2" |
| }, |
| { |
| "dataPath": "params_shard_208.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.36.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "db357c7ddc91b36ecc3898d08f73ef9a" |
| }, |
| { |
| "dataPath": "params_shard_209.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.36.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d27caaf23876294a6b08641e514df9fe" |
| }, |
| { |
| "dataPath": "params_shard_210.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.36.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5302fd6158bd6e8e8ecc5e7ef7585783" |
| }, |
| { |
| "dataPath": "params_shard_211.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.36.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "cf28c0da6e431174b4922798ef146631" |
| }, |
| { |
| "dataPath": "params_shard_212.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.35.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.35.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.36.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.36.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.36.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.36.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.36.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "657edbced6f0304233d54a2488f770bd" |
| }, |
| { |
| "dataPath": "params_shard_213.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.37.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "365de80e6842904128e242887bcb63c1" |
| }, |
| { |
| "dataPath": "params_shard_214.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.37.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "197cfed0308bdb0a2a8fe8856bb4808c" |
| }, |
| { |
| "dataPath": "params_shard_215.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.37.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "17e6728402a4b4a14825cd8384894f24" |
| }, |
| { |
| "dataPath": "params_shard_216.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.37.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1ce9cfbe1da4786cb40e3b60c42910cc" |
| }, |
| { |
| "dataPath": "params_shard_217.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.36.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.36.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.37.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.37.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.37.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.37.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.37.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "23eef08a6f06e25d2ddfdb4f40fe5533" |
| }, |
| { |
| "dataPath": "params_shard_218.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.38.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0ec875928b5cb513469e3d8862984d51" |
| }, |
| { |
| "dataPath": "params_shard_219.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.38.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a63a7110a8f05a4cc2ec39bbb6550e50" |
| }, |
| { |
| "dataPath": "params_shard_220.bin", |
| "format": "raw-shard", |
| "nbytes": 32454656, |
| "records": [ |
| { |
| "name": "model.layers.37.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.37.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.38.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.38.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 32440320 |
| } |
| ], |
| "md5sum": "ed45cf68151f239783da972c9d445d6e" |
| }, |
| { |
| "dataPath": "params_shard_221.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.38.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b7d6e79643e88b4445c4c5f82638baf9" |
| }, |
| { |
| "dataPath": "params_shard_222.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.39.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "dfdc0835ea8581254e71a21532982256" |
| }, |
| { |
| "dataPath": "params_shard_223.bin", |
| "format": "raw-shard", |
| "nbytes": 25917440, |
| "records": [ |
| { |
| "name": "model.layers.38.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.38.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.38.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| }, |
| { |
| "name": "model.layers.38.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 17039360 |
| }, |
| { |
| "name": "model.layers.38.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 17049600 |
| }, |
| { |
| "name": "model.layers.38.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25896960 |
| }, |
| { |
| "name": "model.layers.39.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25907200 |
| } |
| ], |
| "md5sum": "038d06fc8c2a1f0c6cb747414ddc6453" |
| }, |
| { |
| "dataPath": "params_shard_224.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.39.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0f8396963f7e29c7ca4649221823286f" |
| }, |
| { |
| "dataPath": "params_shard_225.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.39.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c74f6faf7677b6344c564502ea4aba07" |
| }, |
| { |
| "dataPath": "params_shard_226.bin", |
| "format": "raw-shard", |
| "nbytes": 28860416, |
| "records": [ |
| { |
| "name": "model.layers.39.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.39.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 8847360 |
| }, |
| { |
| "name": "model.layers.39.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26542080 |
| }, |
| { |
| "name": "model.layers.39.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 26552320 |
| }, |
| { |
| "name": "model.layers.39.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 26566656 |
| } |
| ], |
| "md5sum": "5eaebc63654a86fd08afa457033967c3" |
| }, |
| { |
| "dataPath": "params_shard_227.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.40.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "99d5a4a2101fd611f6dc5c5064f5d2f9" |
| }, |
| { |
| "dataPath": "params_shard_228.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.40.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a7b6edd6d1903613f174d6ab3afd54ea" |
| }, |
| { |
| "dataPath": "params_shard_229.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.40.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "eff69957cdf9b96d13c6b563ddff8360" |
| }, |
| { |
| "dataPath": "params_shard_230.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.40.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "5a13339fbea609c7fb1ecd1997597074" |
| }, |
| { |
| "dataPath": "params_shard_231.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.39.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.39.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.40.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.40.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.40.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.40.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.40.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "7df5db84343583afb2631ef1aece88b5" |
| }, |
| { |
| "dataPath": "params_shard_232.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.41.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ae8366486c03ccdae5a4810f814d575a" |
| }, |
| { |
| "dataPath": "params_shard_233.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.41.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "9b4e22257a46fb934e03daa5cd19823c" |
| }, |
| { |
| "dataPath": "params_shard_234.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.41.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "879e7c98f2967eb4f669a2d60f209510" |
| }, |
| { |
| "dataPath": "params_shard_235.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.41.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "fb3e40d7c7a32f57ac25b23e58baa509" |
| }, |
| { |
| "dataPath": "params_shard_236.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.40.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.40.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.41.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.41.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.41.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.41.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.41.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "5fc4328d2ddbf19a3599c0b34fe5fe9e" |
| }, |
| { |
| "dataPath": "params_shard_237.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.42.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d66778c3fdcc5b14076ddc2ceee82b25" |
| }, |
| { |
| "dataPath": "params_shard_238.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.42.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e86b261a91b28fb2958b90428ec5939f" |
| }, |
| { |
| "dataPath": "params_shard_239.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.42.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ca94cfcf53b088b460eb10c6aa2464a3" |
| }, |
| { |
| "dataPath": "params_shard_240.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.42.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ff5efedc699957e75e6e40322e794193" |
| }, |
| { |
| "dataPath": "params_shard_241.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.41.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.41.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.42.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.42.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.42.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.42.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.42.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "ecd1ade42342780e659ba81a826acbea" |
| }, |
| { |
| "dataPath": "params_shard_242.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.43.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b842910aec127a23445580f9cb1dc5f1" |
| }, |
| { |
| "dataPath": "params_shard_243.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.43.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "34a4b1c7777b1e0078559224a28190fa" |
| }, |
| { |
| "dataPath": "params_shard_244.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.43.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "005470af2463a63d5771720a6ede2f5f" |
| }, |
| { |
| "dataPath": "params_shard_245.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.43.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d72f56fcf87d46a86c4b86d7ec124b5f" |
| }, |
| { |
| "dataPath": "params_shard_246.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.42.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.42.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.43.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.43.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.43.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.43.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.43.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "b8bb65b58858a8f93e47b1eb89ba618e" |
| }, |
| { |
| "dataPath": "params_shard_247.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.44.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "916312107254e3e3b8c2c721263f0d93" |
| }, |
| { |
| "dataPath": "params_shard_248.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.44.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "523f7c092e213906a12a2afcb2f6e2e3" |
| }, |
| { |
| "dataPath": "params_shard_249.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.44.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1e8ae92547e65beb65f317d9752a9840" |
| }, |
| { |
| "dataPath": "params_shard_250.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.44.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "b7f444b0b248008537aa86a03b621b01" |
| }, |
| { |
| "dataPath": "params_shard_251.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.43.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.43.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.44.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.44.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.44.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.44.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.44.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "e030dbcc9ac8a6b8158626fbc758b764" |
| }, |
| { |
| "dataPath": "params_shard_252.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.45.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8e8154d261a87da9c493b72e9ef16b2b" |
| }, |
| { |
| "dataPath": "params_shard_253.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.45.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "dee6506cf26a9850fea161ab80d21afe" |
| }, |
| { |
| "dataPath": "params_shard_254.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.45.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "17ad87cb689cd6c9576b49f60ab5bd22" |
| }, |
| { |
| "dataPath": "params_shard_255.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.45.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c55186da02c6dc96eeee2154907500f1" |
| }, |
| { |
| "dataPath": "params_shard_256.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.44.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.44.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.45.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.45.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.45.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.45.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.45.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "b0403eac163f40d084ccb40ec1e07cba" |
| }, |
| { |
| "dataPath": "params_shard_257.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.46.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "bfb9ef3d6b56f2c0732e8748688ec9d5" |
| }, |
| { |
| "dataPath": "params_shard_258.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.46.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8613f68f520b62954c82f8797790a27f" |
| }, |
| { |
| "dataPath": "params_shard_259.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.46.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "71f3981db6b996a8925481791fd4396c" |
| }, |
| { |
| "dataPath": "params_shard_260.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.46.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ee770a03560ca4618b780f1d2f9923e0" |
| }, |
| { |
| "dataPath": "params_shard_261.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.45.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.45.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.46.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.46.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.46.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.46.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.46.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "1f78325a99034727b200f927d365751c" |
| }, |
| { |
| "dataPath": "params_shard_262.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.47.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8c31ef8ff10be3e08e15d9ef716c1c8a" |
| }, |
| { |
| "dataPath": "params_shard_263.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.47.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "203f90c2ed96dc3b10e3496718f509b7" |
| }, |
| { |
| "dataPath": "params_shard_264.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.47.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c2039f0c23988995936eb05946742804" |
| }, |
| { |
| "dataPath": "params_shard_265.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.47.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "73655722349417f00a89a44dffc24956" |
| }, |
| { |
| "dataPath": "params_shard_266.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.46.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.46.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.47.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.47.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.47.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.47.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.47.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "a412b1a36a17041863d36d79798be8b5" |
| }, |
| { |
| "dataPath": "params_shard_267.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.48.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "fefd2c6f2caef5c6afcaefcb2fb90178" |
| }, |
| { |
| "dataPath": "params_shard_268.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.48.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e44b8570445883229b09fcebfaea2de7" |
| }, |
| { |
| "dataPath": "params_shard_269.bin", |
| "format": "raw-shard", |
| "nbytes": 32454656, |
| "records": [ |
| { |
| "name": "model.layers.47.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.47.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.48.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.48.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 32440320 |
| } |
| ], |
| "md5sum": "981cbcf38b2cca448c2da8dcb44312a1" |
| }, |
| { |
| "dataPath": "params_shard_270.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.48.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1cc941d0896296b23e638f9bf9d62835" |
| }, |
| { |
| "dataPath": "params_shard_271.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.49.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ffd0ed53da77cfda745a9877eb424db4" |
| }, |
| { |
| "dataPath": "params_shard_272.bin", |
| "format": "raw-shard", |
| "nbytes": 25917440, |
| "records": [ |
| { |
| "name": "model.layers.48.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.48.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.48.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| }, |
| { |
| "name": "model.layers.48.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 17039360 |
| }, |
| { |
| "name": "model.layers.48.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 17049600 |
| }, |
| { |
| "name": "model.layers.48.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25896960 |
| }, |
| { |
| "name": "model.layers.49.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 25907200 |
| } |
| ], |
| "md5sum": "c8f4fc64d724933f388fe8f4e2919ccf" |
| }, |
| { |
| "dataPath": "params_shard_273.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.49.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "aab1c1e63403877d1910e46cb5464360" |
| }, |
| { |
| "dataPath": "params_shard_274.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.49.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "076957783010fbae377d76e7ca59dcf4" |
| }, |
| { |
| "dataPath": "params_shard_275.bin", |
| "format": "raw-shard", |
| "nbytes": 28860416, |
| "records": [ |
| { |
| "name": "model.layers.49.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.49.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 8847360 |
| }, |
| { |
| "name": "model.layers.49.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 26542080 |
| }, |
| { |
| "name": "model.layers.49.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 26552320 |
| }, |
| { |
| "name": "model.layers.49.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 26566656 |
| } |
| ], |
| "md5sum": "bd1cb8a1eb4c9364a5cae26873a8350c" |
| }, |
| { |
| "dataPath": "params_shard_276.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.50.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "083cf621ddbb6a726e1cfe450a568be7" |
| }, |
| { |
| "dataPath": "params_shard_277.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.50.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "e1dc1afac08ef83561b720e05a4999c8" |
| }, |
| { |
| "dataPath": "params_shard_278.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.50.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "aca39a9729f32504953418e25cd541b8" |
| }, |
| { |
| "dataPath": "params_shard_279.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.50.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d2a2f07576223fffb622dfed980fd6d6" |
| }, |
| { |
| "dataPath": "params_shard_280.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.49.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.49.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.50.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.50.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.50.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.50.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.50.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "0f1f926632d8d353ec4afac67d6114fc" |
| }, |
| { |
| "dataPath": "params_shard_281.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.51.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "652e74ca4d0af4ed93f03a8adae33ffd" |
| }, |
| { |
| "dataPath": "params_shard_282.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.51.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "7605f555d8fbb548b8008dcf47e80837" |
| }, |
| { |
| "dataPath": "params_shard_283.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.51.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f68a91bf78eb211a5d4ae96518ac1d89" |
| }, |
| { |
| "dataPath": "params_shard_284.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.51.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "db8f6658837f97cf41e1dd298496e3d7" |
| }, |
| { |
| "dataPath": "params_shard_285.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.50.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.50.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.51.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.51.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.51.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.51.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.51.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "20008bc883ecdad91ab87b4451b712b4" |
| }, |
| { |
| "dataPath": "params_shard_286.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.52.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "6a566bdfbb25df6bf6f07e70e565869a" |
| }, |
| { |
| "dataPath": "params_shard_287.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.52.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8f3ba13fff26573ece27eb377a03b540" |
| }, |
| { |
| "dataPath": "params_shard_288.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.52.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "ebc8cd90647d684d91bc4eb3ed45bd8e" |
| }, |
| { |
| "dataPath": "params_shard_289.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.52.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "4d3d80cc1d2c342bae6e47c12274a0ad" |
| }, |
| { |
| "dataPath": "params_shard_290.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.51.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.51.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.52.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.52.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.52.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.52.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.52.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "88994ddaf32665bb24032055b77bf175" |
| }, |
| { |
| "dataPath": "params_shard_291.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.53.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0f702736f4c49fa30ea53f9669e698d2" |
| }, |
| { |
| "dataPath": "params_shard_292.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.53.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "299fea6ff0668ab003b74819ed6cde10" |
| }, |
| { |
| "dataPath": "params_shard_293.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.53.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "61fc3bb1e3bfcc9a6633bc3dff4df1af" |
| }, |
| { |
| "dataPath": "params_shard_294.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.53.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "1dc9bf9dd138ff308f30eb9b2ed095a4" |
| }, |
| { |
| "dataPath": "params_shard_295.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.52.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.52.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.53.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.53.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.53.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.53.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.53.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "b02e8d0d8eace3e97e5886a2e626f8f9" |
| }, |
| { |
| "dataPath": "params_shard_296.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.54.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d255bf8f856150697a4d88bf9cc59dca" |
| }, |
| { |
| "dataPath": "params_shard_297.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.54.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "806e3bfc2daae8313858521fc67b49bf" |
| }, |
| { |
| "dataPath": "params_shard_298.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.54.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "61f086359f7a8c3096a1236e6c96e695" |
| }, |
| { |
| "dataPath": "params_shard_299.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.54.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "62637f0827b0ed708e1670926c154d58" |
| }, |
| { |
| "dataPath": "params_shard_300.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.53.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.53.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.54.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.54.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.54.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.54.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.54.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "ccee7b843acb179bafe3ce4b56073513" |
| }, |
| { |
| "dataPath": "params_shard_301.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.55.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "a61a6457466e70865bbd718d81e0289b" |
| }, |
| { |
| "dataPath": "params_shard_302.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.55.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "67ea2edba0fdf52d72a136acdfb89318" |
| }, |
| { |
| "dataPath": "params_shard_303.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.55.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "0b536e4a6565484f8997dca4657deb2f" |
| }, |
| { |
| "dataPath": "params_shard_304.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.55.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c69455abd06eeb0c82603d74e90fd4e5" |
| }, |
| { |
| "dataPath": "params_shard_305.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.54.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.54.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.55.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.55.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.55.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.55.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.55.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "a65fa8fc98ac773ae362ade852f52b01" |
| }, |
| { |
| "dataPath": "params_shard_306.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.56.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "d9976209e95867594d9f4b6ae74719f9" |
| }, |
| { |
| "dataPath": "params_shard_307.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.56.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "c99fd80e3dccf9f127b21a425de8bf11" |
| }, |
| { |
| "dataPath": "params_shard_308.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.56.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "f67be28a4d18a2234d92411bc4af8583" |
| }, |
| { |
| "dataPath": "params_shard_309.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.56.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "be34db475368f030b61b77f4017322a3" |
| }, |
| { |
| "dataPath": "params_shard_310.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.55.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.55.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.56.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.56.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.56.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.56.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.56.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "93e2bea64ca0995e7d282e789cbe1b52" |
| }, |
| { |
| "dataPath": "params_shard_311.bin", |
| "format": "raw-shard", |
| "nbytes": 70778880, |
| "records": [ |
| { |
| "name": "model.layers.57.mlp.down_proj.q_weight", |
| "shape": [ |
| 5120, |
| 3456 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 70778880, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "71a2e8e92407536a1c935dbd37d24328" |
| }, |
| { |
| "dataPath": "params_shard_312.bin", |
| "format": "raw-shard", |
| "nbytes": 141557760, |
| "records": [ |
| { |
| "name": "model.layers.57.mlp.gate_up_proj.q_weight", |
| "shape": [ |
| 55296, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 141557760, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "567316fa2f5d43a5b99ed8bed261fde7" |
| }, |
| { |
| "dataPath": "params_shard_313.bin", |
| "format": "raw-shard", |
| "nbytes": 17694720, |
| "records": [ |
| { |
| "name": "model.layers.57.mlp.gate_up_proj.q_scale", |
| "shape": [ |
| 55296, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 17694720, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "dce942ffd01896b2fb62fb37d8d88d06" |
| }, |
| { |
| "dataPath": "params_shard_314.bin", |
| "format": "raw-shard", |
| "nbytes": 18350080, |
| "records": [ |
| { |
| "name": "model.layers.57.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 0 |
| } |
| ], |
| "md5sum": "8a52ca05a4c0022e27351f712d9c846b" |
| }, |
| { |
| "dataPath": "params_shard_315.bin", |
| "format": "raw-shard", |
| "nbytes": 25921536, |
| "records": [ |
| { |
| "name": "model.layers.56.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.56.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.57.input_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.57.mlp.down_proj.q_scale", |
| "shape": [ |
| 5120, |
| 864 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 8847360, |
| "byteOffset": 14755840 |
| }, |
| { |
| "name": "model.layers.57.post_attention_layernorm.weight", |
| "shape": [ |
| 5120 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 10240, |
| "byteOffset": 23603200 |
| }, |
| { |
| "name": "model.layers.57.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 23613440 |
| }, |
| { |
| "name": "model.layers.57.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 23627776 |
| } |
| ], |
| "md5sum": "55f390c7fe00811328e6152ee1dbbe89" |
| }, |
| { |
| "dataPath": "params_shard_316.bin", |
| "format": "raw-shard", |
| "nbytes": 33110016, |
| "records": [ |
| { |
| "name": "model.layers.57.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.57.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 13107200 |
| }, |
| { |
| "name": "model.layers.58.self_attn.c_attn.bias", |
| "shape": [ |
| 7168 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 14336, |
| "byteOffset": 14745600 |
| }, |
| { |
| "name": "model.layers.58.self_attn.c_attn.q_weight", |
| "shape": [ |
| 7168, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 18350080, |
| "byteOffset": 14759936 |
| } |
| ], |
| "md5sum": "b932a3b7fbc1d97edfefe1861e6a3684" |
| }, |
| { |
| "dataPath": "params_shard_317.bin", |
| "format": "raw-shard", |
| "nbytes": 17039360, |
| "records": [ |
| { |
| "name": "model.layers.58.self_attn.c_attn.q_scale", |
| "shape": [ |
| 7168, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 2293760, |
| "byteOffset": 0 |
| }, |
| { |
| "name": "model.layers.58.self_attn.o_proj.q_weight", |
| "shape": [ |
| 5120, |
| 640 |
| ], |
| "dtype": "uint32", |
| "format": "f32-to-bf16", |
| "nbytes": 13107200, |
| "byteOffset": 2293760 |
| }, |
| { |
| "name": "model.layers.58.self_attn.o_proj.q_scale", |
| "shape": [ |
| 5120, |
| 160 |
| ], |
| "dtype": "float16", |
| "format": "f32-to-bf16", |
| "nbytes": 1638400, |
| "byteOffset": 15400960 |
| } |
| ], |
| "md5sum": "e85f2066e29d7e2332cd3920f83d1652" |
| } |
| ] |
| } |
