diff --git "a/scores/gemma-4-26B-A4B-it-Q2_K.md" "b/scores/gemma-4-26B-A4B-it-Q2_K.md" new file mode 100644--- /dev/null +++ "b/scores/gemma-4-26B-A4B-it-Q2_K.md" @@ -0,0 +1,1742 @@ +# gemma-4-26B-A4B-it-Q2_K.gguf - GGUF Internal File Dump + +- Endian: LITTLE endian + +## Key Value Metadata Store + +There are 57 key-value pairs in this file + +| POS | TYPE | Count | Key | Value | +| ---: | :-------- | -----: | :-------------------------------------- | :--------------------------------------------------------------------------------------------------------------- | +| 1 | UINT32 | 1 | GGUF.version | 3 | +| 2 | UINT64 | 1 | GGUF.tensor_count | 658 | +| 3 | UINT64 | 1 | GGUF.kv_count | 54 | +| 4 | STRING | 1 | general.architecture | `gemma4` | +| 5 | STRING | 1 | general.type | `model` | +| 6 | INT32 | 1 | general.sampling.top_k | 64 | +| 7 | FLOAT32 | 1 | general.sampling.top_p | 0.95 | +| 8 | FLOAT32 | 1 | general.sampling.temp | 1.0 | +| 9 | STRING | 1 | general.name | `Gemma 4 26B A4B It` | +| 10 | STRING | 1 | general.finetune | `it` | +| 11 | STRING | 1 | general.basename | `gemma-4` | +| 12 | STRING | 1 | general.size_label | `26B-A4B` | +| 13 | STRING | 1 | general.license | `apache-2.0` | +| 14 | STRING | 1 | general.license.link | `https://ai.google.dev/gemma/docs/gemma_4_license` | +| 15 | [STRING] | 1 | general.tags | [ `image-text-to-text` ] | +| 16 | UINT32 | 1 | gemma4.block_count | 30 | +| 17 | UINT32 | 1 | gemma4.context_length | 262144 | +| 18 | UINT32 | 1 | gemma4.embedding_length | 2816 | +| 19 | UINT32 | 1 | gemma4.feed_forward_length | 2112 | +| 20 | UINT32 | 1 | gemma4.attention.head_count | 16 | +| 21 | [INT32] | 30 | gemma4.attention.head_count_kv | [ 8, 8, 8, 8, 8, 2, 8, ... ] | +| 22 | FLOAT32 | 1 | gemma4.rope.freq_base | 1e+06 | +| 23 | FLOAT32 | 1 | gemma4.rope.freq_base_swa | 10000.0 | +| 24 | FLOAT32 | 1 | gemma4.attention.layer_norm_rms_epsilon | 1e-06 | +| 25 | UINT32 | 1 | gemma4.expert_count | 128 | +| 26 | UINT32 | 1 | gemma4.expert_used_count | 8 | +| 27 | UINT32 | 1 | gemma4.attention.key_length | 512 | +| 28 | UINT32 | 1 | gemma4.attention.value_length | 512 | +| 29 | FLOAT32 | 1 | gemma4.final_logit_softcapping | 30.0 | +| 30 | UINT32 | 1 | gemma4.attention.sliding_window | 1024 | +| 31 | UINT32 | 1 | gemma4.attention.shared_kv_layers | 0 | +| 32 | UINT32 | 1 | gemma4.embedding_length_per_layer_input | 0 | +| 33 | [BOOL] | 30 | gemma4.attention.sliding_window_pattern | [ True, True, True, True, True, False, True, ... ] | +| 34 | UINT32 | 1 | gemma4.attention.key_length_swa | 256 | +| 35 | UINT32 | 1 | gemma4.attention.value_length_swa | 256 | +| 36 | UINT32 | 1 | gemma4.expert_feed_forward_length | 704 | +| 37 | UINT32 | 1 | gemma4.rope.dimension_count | 512 | +| 38 | UINT32 | 1 | gemma4.rope.dimension_count_swa | 256 | +| 39 | STRING | 1 | tokenizer.ggml.model | `gemma4` | +| 40 | [STRING] | 262144 | tokenizer.ggml.tokens | [ ``, ``, ``, ``, ``, ... ] | +| 41 | [FLOAT32] | 262144 | tokenizer.ggml.scores | [ -1000.0, -1000.0, -1000.0, -1000.0, -1000.0, -1000.0, -1000.0, ... ] | +| 42 | [INT32] | 262144 | tokenizer.ggml.token_type | [ 3, 3, 3, 3, 3, 1, 1, ... ] | +| 45 | [STRING] | 514906 | tokenizer.ggml.merges | [ ``...``, `▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁`...`▁▁▁▁▁▁▁▁▁▁▁▁▁ ▁`, ``...``, ``...``, `▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁`...`▁▁▁▁▁▁▁▁▁▁▁▁ ▁▁`, ... ] | +| 44 | UINT32 | 1 | tokenizer.ggml.bos_token_id | 2 | +| 45 | UINT32 | 1 | tokenizer.ggml.eos_token_id | 1 | +| 46 | UINT32 | 1 | tokenizer.ggml.unknown_token_id | 3 | +| 47 | UINT32 | 1 | tokenizer.ggml.padding_token_id | 0 | +| 48 | UINT32 | 1 | tokenizer.ggml.mask_token_id | 4 | +| 49 | STRING | 1 | tokenizer.chat_template | `{%- macro format_parameters(pr`...` {%- endif -%} {%- endif -%}` | +| 50 | BOOL | 1 | tokenizer.ggml.add_space_prefix | False | +| 51 | BOOL | 1 | tokenizer.ggml.add_bos_token | True | +| 52 | UINT32 | 1 | general.quantization_version | 2 | +| 53 | UINT32 | 1 | general.file_type | 28 | +| 54 | STRING | 1 | quantize.imatrix.file | `gemma-4-26B-A4B-it-WIP/imatrix`...`gemma-4-26B-A4B-it-medium.gguf` | +| 55 | STRING | 1 | quantize.imatrix.dataset | `../datasets/imatrix/combined_eur_medium.txt` | +| 56 | UINT32 | 1 | quantize.imatrix.entries_count | 295 | +| 57 | UINT32 | 1 | quantize.imatrix.chunks_count | 2471 | + +## Tensors Overview ~25B Elements + +Total number of elements in all tensors: 25233142046 Elements + +- [/Users/ed/Development/AI/hf/gemma-4-26B-A4B-it-WIP/gemma-4-26B-A4B-it-Q2\_K.gguf - GGUF Internal File Dump](#userseddevelopmentaihfgemma-4-26b-a4b-it-wipgemma-4-26b-a4b-it-q2_kgguf---gguf-internal-file-dump) + - [Key Value Metadata Store](#key-value-metadata-store) + - [Tensors Overview ~25B Elements](#tensors-overview-25b-elements) + - [Tensor Data Offset](#tensor-data-offset) + - [Base Tensor Group : ~738M Elements](#base-tensor-group--738m-elements) + - [Block 0 Tensor Group : ~814M Elements](#block-0-tensor-group--814m-elements) + - [Block 1 Tensor Group : ~814M Elements](#block-1-tensor-group--814m-elements) + - [Block 2 Tensor Group : ~814M Elements](#block-2-tensor-group--814m-elements) + - [Block 3 Tensor Group : ~814M Elements](#block-3-tensor-group--814m-elements) + - [Block 4 Tensor Group : ~814M Elements](#block-4-tensor-group--814m-elements) + - [Block 5 Tensor Group : ~829M Elements](#block-5-tensor-group--829m-elements) + - [Block 6 Tensor Group : ~814M Elements](#block-6-tensor-group--814m-elements) + - [Block 7 Tensor Group : ~814M Elements](#block-7-tensor-group--814m-elements) + - [Block 8 Tensor Group : ~814M Elements](#block-8-tensor-group--814m-elements) + - [Block 9 Tensor Group : ~814M Elements](#block-9-tensor-group--814m-elements) + - [Block 10 Tensor Group : ~814M Elements](#block-10-tensor-group--814m-elements) + - [Block 11 Tensor Group : ~829M Elements](#block-11-tensor-group--829m-elements) + - [Block 12 Tensor Group : ~814M Elements](#block-12-tensor-group--814m-elements) + - [Block 13 Tensor Group : ~814M Elements](#block-13-tensor-group--814m-elements) + - [Block 14 Tensor Group : ~814M Elements](#block-14-tensor-group--814m-elements) + - [Block 15 Tensor Group : ~814M Elements](#block-15-tensor-group--814m-elements) + - [Block 16 Tensor Group : ~814M Elements](#block-16-tensor-group--814m-elements) + - [Block 17 Tensor Group : ~829M Elements](#block-17-tensor-group--829m-elements) + - [Block 18 Tensor Group : ~814M Elements](#block-18-tensor-group--814m-elements) + - [Block 19 Tensor Group : ~814M Elements](#block-19-tensor-group--814m-elements) + - [Block 20 Tensor Group : ~814M Elements](#block-20-tensor-group--814m-elements) + - [Block 21 Tensor Group : ~814M Elements](#block-21-tensor-group--814m-elements) + - [Block 22 Tensor Group : ~814M Elements](#block-22-tensor-group--814m-elements) + - [Block 23 Tensor Group : ~829M Elements](#block-23-tensor-group--829m-elements) + - [Block 24 Tensor Group : ~814M Elements](#block-24-tensor-group--814m-elements) + - [Block 25 Tensor Group : ~814M Elements](#block-25-tensor-group--814m-elements) + - [Block 26 Tensor Group : ~814M Elements](#block-26-tensor-group--814m-elements) + - [Block 27 Tensor Group : ~814M Elements](#block-27-tensor-group--814m-elements) + - [Block 28 Tensor Group : ~814M Elements](#block-28-tensor-group--814m-elements) + - [Block 29 Tensor Group : ~829M Elements](#block-29-tensor-group--829m-elements) + +### Tensor Data Offset + +This table contains the offset and data segment relative to start of file + +| T_ID | Tensor Layer Name | Data Offset (B) | Data Size (B) | +| ---: | :-------------------------------- | --------------: | ------------: | +| 0 | output_norm.weight | 0xf16c00 | 0x2c00 | +| 1 | rope_freqs.weight | 0xf19800 | 0x400 | +| 2 | token_embd.weight | 0xf19c00 | 0xe700000 | +| 3 | blk.0.attn_k.weight | 0xf619c00 | 0x113000 | +| 4 | blk.0.attn_k_norm.weight | 0xf72cc00 | 0x400 | +| 5 | blk.0.attn_norm.weight | 0xf72d000 | 0x2c00 | +| 6 | blk.0.attn_output.weight | 0xf72fc00 | 0x226000 | +| 7 | blk.0.attn_q.weight | 0xf955c00 | 0x226000 | +| 8 | blk.0.attn_q_norm.weight | 0xfb7bc00 | 0x400 | +| 9 | blk.0.attn_v.weight | 0xfb7c000 | 0x113000 | +| 10 | blk.0.ffn_down.weight | 0xfc8f000 | 0x330c00 | +| 11 | blk.0.ffn_down_exps.scale | 0xffbfc00 | 0x200 | +| 12 | blk.0.ffn_down_exps.weight | 0xffbfe00 | 0x8820000 | +| 13 | blk.0.ffn_gate.weight | 0x187dfe00 | 0x11b980 | +| 14 | blk.0.ffn_gate_inp.scale | 0x188fb780 | 0x2c00 | +| 15 | blk.0.ffn_gate_inp.weight | 0x188fe380 | 0x160000 | +| 16 | blk.0.ffn_gate_up_exps.weight | 0x18a5e380 | 0x5e88000 | +| 17 | blk.0.ffn_norm.weight | 0x1e8e6380 | 0x2c00 | +| 18 | blk.0.ffn_up.weight | 0x1e8e8f80 | 0x11b980 | +| 19 | blk.0.layer_output_scale.weight | 0x1ea04900 | 0x4 | +| 20 | blk.0.post_attention_norm.weight | 0x1ea04920 | 0x2c00 | +| 21 | blk.0.post_ffw_norm.weight | 0x1ea07520 | 0x2c00 | +| 22 | blk.0.post_ffw_norm_1.weight | 0x1ea0a120 | 0x2c00 | +| 23 | blk.0.post_ffw_norm_2.weight | 0x1ea0cd20 | 0x2c00 | +| 24 | blk.0.pre_ffw_norm_2.weight | 0x1ea0f920 | 0x2c00 | +| 25 | blk.1.attn_k.weight | 0x1ea12520 | 0x113000 | +| 26 | blk.1.attn_k_norm.weight | 0x1eb25520 | 0x400 | +| 27 | blk.1.attn_norm.weight | 0x1eb25920 | 0x2c00 | +| 28 | blk.1.attn_output.weight | 0x1eb28520 | 0x226000 | +| 29 | blk.1.attn_q.weight | 0x1ed4e520 | 0x226000 | +| 30 | blk.1.attn_q_norm.weight | 0x1ef74520 | 0x400 | +| 31 | blk.1.attn_v.weight | 0x1ef74920 | 0x113000 | +| 32 | blk.1.ffn_down.weight | 0x1f087920 | 0x330c00 | +| 33 | blk.1.ffn_down_exps.scale | 0x1f3b8520 | 0x200 | +| 34 | blk.1.ffn_down_exps.weight | 0x1f3b8720 | 0x8820000 | +| 35 | blk.1.ffn_gate.weight | 0x27bd8720 | 0x11b980 | +| 36 | blk.1.ffn_gate_inp.scale | 0x27cf40a0 | 0x2c00 | +| 37 | blk.1.ffn_gate_inp.weight | 0x27cf6ca0 | 0x160000 | +| 38 | blk.1.ffn_gate_up_exps.weight | 0x27e56ca0 | 0x5e88000 | +| 39 | blk.1.ffn_norm.weight | 0x2dcdeca0 | 0x2c00 | +| 40 | blk.1.ffn_up.weight | 0x2dce18a0 | 0x11b980 | +| 41 | blk.1.layer_output_scale.weight | 0x2ddfd220 | 0x4 | +| 42 | blk.1.post_attention_norm.weight | 0x2ddfd240 | 0x2c00 | +| 43 | blk.1.post_ffw_norm.weight | 0x2ddffe40 | 0x2c00 | +| 44 | blk.1.post_ffw_norm_1.weight | 0x2de02a40 | 0x2c00 | +| 45 | blk.1.post_ffw_norm_2.weight | 0x2de05640 | 0x2c00 | +| 46 | blk.1.pre_ffw_norm_2.weight | 0x2de08240 | 0x2c00 | +| 47 | blk.2.attn_k.weight | 0x2de0ae40 | 0x113000 | +| 48 | blk.2.attn_k_norm.weight | 0x2df1de40 | 0x400 | +| 49 | blk.2.attn_norm.weight | 0x2df1e240 | 0x2c00 | +| 50 | blk.2.attn_output.weight | 0x2df20e40 | 0x226000 | +| 51 | blk.2.attn_q.weight | 0x2e146e40 | 0x226000 | +| 52 | blk.2.attn_q_norm.weight | 0x2e36ce40 | 0x400 | +| 53 | blk.2.attn_v.weight | 0x2e36d240 | 0x113000 | +| 54 | blk.2.ffn_down.weight | 0x2e480240 | 0x330c00 | +| 55 | blk.2.ffn_down_exps.scale | 0x2e7b0e40 | 0x200 | +| 56 | blk.2.ffn_down_exps.weight | 0x2e7b1040 | 0x8820000 | +| 57 | blk.2.ffn_gate.weight | 0x36fd1040 | 0x11b980 | +| 58 | blk.2.ffn_gate_inp.scale | 0x370ec9c0 | 0x2c00 | +| 59 | blk.2.ffn_gate_inp.weight | 0x370ef5c0 | 0x160000 | +| 60 | blk.2.ffn_gate_up_exps.weight | 0x3724f5c0 | 0x5e88000 | +| 61 | blk.2.ffn_norm.weight | 0x3d0d75c0 | 0x2c00 | +| 62 | blk.2.ffn_up.weight | 0x3d0da1c0 | 0x11b980 | +| 63 | blk.2.layer_output_scale.weight | 0x3d1f5b40 | 0x4 | +| 64 | blk.2.post_attention_norm.weight | 0x3d1f5b60 | 0x2c00 | +| 65 | blk.2.post_ffw_norm.weight | 0x3d1f8760 | 0x2c00 | +| 66 | blk.2.post_ffw_norm_1.weight | 0x3d1fb360 | 0x2c00 | +| 67 | blk.2.post_ffw_norm_2.weight | 0x3d1fdf60 | 0x2c00 | +| 68 | blk.2.pre_ffw_norm_2.weight | 0x3d200b60 | 0x2c00 | +| 69 | blk.3.attn_k.weight | 0x3d203760 | 0x113000 | +| 70 | blk.3.attn_k_norm.weight | 0x3d316760 | 0x400 | +| 71 | blk.3.attn_norm.weight | 0x3d316b60 | 0x2c00 | +| 72 | blk.3.attn_output.weight | 0x3d319760 | 0x226000 | +| 73 | blk.3.attn_q.weight | 0x3d53f760 | 0x226000 | +| 74 | blk.3.attn_q_norm.weight | 0x3d765760 | 0x400 | +| 75 | blk.3.attn_v.weight | 0x3d765b60 | 0x113000 | +| 76 | blk.3.ffn_down.weight | 0x3d878b60 | 0x330c00 | +| 77 | blk.3.ffn_down_exps.scale | 0x3dba9760 | 0x200 | +| 78 | blk.3.ffn_down_exps.weight | 0x3dba9960 | 0x8820000 | +| 79 | blk.3.ffn_gate.weight | 0x463c9960 | 0x11b980 | +| 80 | blk.3.ffn_gate_inp.scale | 0x464e52e0 | 0x2c00 | +| 81 | blk.3.ffn_gate_inp.weight | 0x464e7ee0 | 0x160000 | +| 82 | blk.3.ffn_gate_up_exps.weight | 0x46647ee0 | 0x5e88000 | +| 83 | blk.3.ffn_norm.weight | 0x4c4cfee0 | 0x2c00 | +| 84 | blk.3.ffn_up.weight | 0x4c4d2ae0 | 0x11b980 | +| 85 | blk.3.layer_output_scale.weight | 0x4c5ee460 | 0x4 | +| 86 | blk.3.post_attention_norm.weight | 0x4c5ee480 | 0x2c00 | +| 87 | blk.3.post_ffw_norm.weight | 0x4c5f1080 | 0x2c00 | +| 88 | blk.3.post_ffw_norm_1.weight | 0x4c5f3c80 | 0x2c00 | +| 89 | blk.3.post_ffw_norm_2.weight | 0x4c5f6880 | 0x2c00 | +| 90 | blk.3.pre_ffw_norm_2.weight | 0x4c5f9480 | 0x2c00 | +| 91 | blk.4.attn_k.weight | 0x4c5fc080 | 0x113000 | +| 92 | blk.4.attn_k_norm.weight | 0x4c70f080 | 0x400 | +| 93 | blk.4.attn_norm.weight | 0x4c70f480 | 0x2c00 | +| 94 | blk.4.attn_output.weight | 0x4c712080 | 0x226000 | +| 95 | blk.4.attn_q.weight | 0x4c938080 | 0x226000 | +| 96 | blk.4.attn_q_norm.weight | 0x4cb5e080 | 0x400 | +| 97 | blk.4.attn_v.weight | 0x4cb5e480 | 0x113000 | +| 98 | blk.4.ffn_down.weight | 0x4cc71480 | 0x330c00 | +| 99 | blk.4.ffn_down_exps.scale | 0x4cfa2080 | 0x200 | +| 100 | blk.4.ffn_down_exps.weight | 0x4cfa2280 | 0x8820000 | +| 101 | blk.4.ffn_gate.weight | 0x557c2280 | 0x11b980 | +| 102 | blk.4.ffn_gate_inp.scale | 0x558ddc00 | 0x2c00 | +| 103 | blk.4.ffn_gate_inp.weight | 0x558e0800 | 0x160000 | +| 104 | blk.4.ffn_gate_up_exps.weight | 0x55a40800 | 0x5e88000 | +| 105 | blk.4.ffn_norm.weight | 0x5b8c8800 | 0x2c00 | +| 106 | blk.4.ffn_up.weight | 0x5b8cb400 | 0x11b980 | +| 107 | blk.4.layer_output_scale.weight | 0x5b9e6d80 | 0x4 | +| 108 | blk.4.post_attention_norm.weight | 0x5b9e6da0 | 0x2c00 | +| 109 | blk.4.post_ffw_norm.weight | 0x5b9e99a0 | 0x2c00 | +| 110 | blk.4.post_ffw_norm_1.weight | 0x5b9ec5a0 | 0x2c00 | +| 111 | blk.4.post_ffw_norm_2.weight | 0x5b9ef1a0 | 0x2c00 | +| 112 | blk.4.pre_ffw_norm_2.weight | 0x5b9f1da0 | 0x2c00 | +| 113 | blk.5.attn_k.weight | 0x5b9f49a0 | 0x89800 | +| 114 | blk.5.attn_k_norm.weight | 0x5ba7e1a0 | 0x800 | +| 115 | blk.5.attn_norm.weight | 0x5ba7e9a0 | 0x2c00 | +| 116 | blk.5.attn_output.weight | 0x5ba815a0 | 0x44c000 | +| 117 | blk.5.attn_q.weight | 0x5becd5a0 | 0x44c000 | +| 118 | blk.5.attn_q_norm.weight | 0x5c3195a0 | 0x800 | +| 119 | blk.5.ffn_down.weight | 0x5c319da0 | 0x330c00 | +| 120 | blk.5.ffn_down_exps.scale | 0x5c64a9a0 | 0x200 | +| 121 | blk.5.ffn_down_exps.weight | 0x5c64aba0 | 0x8820000 | +| 122 | blk.5.ffn_gate.weight | 0x64e6aba0 | 0x11b980 | +| 123 | blk.5.ffn_gate_inp.scale | 0x64f86520 | 0x2c00 | +| 124 | blk.5.ffn_gate_inp.weight | 0x64f89120 | 0x160000 | +| 125 | blk.5.ffn_gate_up_exps.weight | 0x650e9120 | 0x5e88000 | +| 126 | blk.5.ffn_norm.weight | 0x6af71120 | 0x2c00 | +| 127 | blk.5.ffn_up.weight | 0x6af73d20 | 0x11b980 | +| 128 | blk.5.layer_output_scale.weight | 0x6b08f6a0 | 0x4 | +| 129 | blk.5.post_attention_norm.weight | 0x6b08f6c0 | 0x2c00 | +| 130 | blk.5.post_ffw_norm.weight | 0x6b0922c0 | 0x2c00 | +| 131 | blk.5.post_ffw_norm_1.weight | 0x6b094ec0 | 0x2c00 | +| 132 | blk.5.post_ffw_norm_2.weight | 0x6b097ac0 | 0x2c00 | +| 133 | blk.5.pre_ffw_norm_2.weight | 0x6b09a6c0 | 0x2c00 | +| 134 | blk.6.attn_k.weight | 0x6b09d2c0 | 0x113000 | +| 135 | blk.6.attn_k_norm.weight | 0x6b1b02c0 | 0x400 | +| 136 | blk.6.attn_norm.weight | 0x6b1b06c0 | 0x2c00 | +| 137 | blk.6.attn_output.weight | 0x6b1b32c0 | 0x226000 | +| 138 | blk.6.attn_q.weight | 0x6b3d92c0 | 0x226000 | +| 139 | blk.6.attn_q_norm.weight | 0x6b5ff2c0 | 0x400 | +| 140 | blk.6.attn_v.weight | 0x6b5ff6c0 | 0x113000 | +| 141 | blk.6.ffn_down.weight | 0x6b7126c0 | 0x330c00 | +| 142 | blk.6.ffn_down_exps.scale | 0x6ba432c0 | 0x200 | +| 143 | blk.6.ffn_down_exps.weight | 0x6ba434c0 | 0x8820000 | +| 144 | blk.6.ffn_gate.weight | 0x742634c0 | 0x11b980 | +| 145 | blk.6.ffn_gate_inp.scale | 0x7437ee40 | 0x2c00 | +| 146 | blk.6.ffn_gate_inp.weight | 0x74381a40 | 0x160000 | +| 147 | blk.6.ffn_gate_up_exps.weight | 0x744e1a40 | 0x5e88000 | +| 148 | blk.6.ffn_norm.weight | 0x7a369a40 | 0x2c00 | +| 149 | blk.6.ffn_up.weight | 0x7a36c640 | 0x11b980 | +| 150 | blk.6.layer_output_scale.weight | 0x7a487fc0 | 0x4 | +| 151 | blk.6.post_attention_norm.weight | 0x7a487fe0 | 0x2c00 | +| 152 | blk.6.post_ffw_norm.weight | 0x7a48abe0 | 0x2c00 | +| 153 | blk.6.post_ffw_norm_1.weight | 0x7a48d7e0 | 0x2c00 | +| 154 | blk.6.post_ffw_norm_2.weight | 0x7a4903e0 | 0x2c00 | +| 155 | blk.6.pre_ffw_norm_2.weight | 0x7a492fe0 | 0x2c00 | +| 156 | blk.7.attn_k.weight | 0x7a495be0 | 0x113000 | +| 157 | blk.7.attn_k_norm.weight | 0x7a5a8be0 | 0x400 | +| 158 | blk.7.attn_norm.weight | 0x7a5a8fe0 | 0x2c00 | +| 159 | blk.7.attn_output.weight | 0x7a5abbe0 | 0x226000 | +| 160 | blk.7.attn_q.weight | 0x7a7d1be0 | 0x226000 | +| 161 | blk.7.attn_q_norm.weight | 0x7a9f7be0 | 0x400 | +| 162 | blk.7.attn_v.weight | 0x7a9f7fe0 | 0x113000 | +| 163 | blk.7.ffn_down.weight | 0x7ab0afe0 | 0x330c00 | +| 164 | blk.7.ffn_down_exps.scale | 0x7ae3bbe0 | 0x200 | +| 165 | blk.7.ffn_down_exps.weight | 0x7ae3bde0 | 0x8820000 | +| 166 | blk.7.ffn_gate.weight | 0x8365bde0 | 0x11b980 | +| 167 | blk.7.ffn_gate_inp.scale | 0x83777760 | 0x2c00 | +| 168 | blk.7.ffn_gate_inp.weight | 0x8377a360 | 0x160000 | +| 169 | blk.7.ffn_gate_up_exps.weight | 0x838da360 | 0x5e88000 | +| 170 | blk.7.ffn_norm.weight | 0x89762360 | 0x2c00 | +| 171 | blk.7.ffn_up.weight | 0x89764f60 | 0x11b980 | +| 172 | blk.7.layer_output_scale.weight | 0x898808e0 | 0x4 | +| 173 | blk.7.post_attention_norm.weight | 0x89880900 | 0x2c00 | +| 174 | blk.7.post_ffw_norm.weight | 0x89883500 | 0x2c00 | +| 175 | blk.7.post_ffw_norm_1.weight | 0x89886100 | 0x2c00 | +| 176 | blk.7.post_ffw_norm_2.weight | 0x89888d00 | 0x2c00 | +| 177 | blk.7.pre_ffw_norm_2.weight | 0x8988b900 | 0x2c00 | +| 178 | blk.8.attn_k.weight | 0x8988e500 | 0x113000 | +| 179 | blk.8.attn_k_norm.weight | 0x899a1500 | 0x400 | +| 180 | blk.8.attn_norm.weight | 0x899a1900 | 0x2c00 | +| 181 | blk.8.attn_output.weight | 0x899a4500 | 0x226000 | +| 182 | blk.8.attn_q.weight | 0x89bca500 | 0x226000 | +| 183 | blk.8.attn_q_norm.weight | 0x89df0500 | 0x400 | +| 184 | blk.8.attn_v.weight | 0x89df0900 | 0x113000 | +| 185 | blk.8.ffn_down.weight | 0x89f03900 | 0x330c00 | +| 186 | blk.8.ffn_down_exps.scale | 0x8a234500 | 0x200 | +| 187 | blk.8.ffn_down_exps.weight | 0x8a234700 | 0x8820000 | +| 188 | blk.8.ffn_gate.weight | 0x92a54700 | 0x11b980 | +| 189 | blk.8.ffn_gate_inp.scale | 0x92b70080 | 0x2c00 | +| 190 | blk.8.ffn_gate_inp.weight | 0x92b72c80 | 0x160000 | +| 191 | blk.8.ffn_gate_up_exps.weight | 0x92cd2c80 | 0x5e88000 | +| 192 | blk.8.ffn_norm.weight | 0x98b5ac80 | 0x2c00 | +| 193 | blk.8.ffn_up.weight | 0x98b5d880 | 0x11b980 | +| 194 | blk.8.layer_output_scale.weight | 0x98c79200 | 0x4 | +| 195 | blk.8.post_attention_norm.weight | 0x98c79220 | 0x2c00 | +| 196 | blk.8.post_ffw_norm.weight | 0x98c7be20 | 0x2c00 | +| 197 | blk.8.post_ffw_norm_1.weight | 0x98c7ea20 | 0x2c00 | +| 198 | blk.8.post_ffw_norm_2.weight | 0x98c81620 | 0x2c00 | +| 199 | blk.8.pre_ffw_norm_2.weight | 0x98c84220 | 0x2c00 | +| 200 | blk.9.attn_k.weight | 0x98c86e20 | 0x113000 | +| 201 | blk.9.attn_k_norm.weight | 0x98d99e20 | 0x400 | +| 202 | blk.9.attn_norm.weight | 0x98d9a220 | 0x2c00 | +| 203 | blk.9.attn_output.weight | 0x98d9ce20 | 0x226000 | +| 204 | blk.9.attn_q.weight | 0x98fc2e20 | 0x226000 | +| 205 | blk.9.attn_q_norm.weight | 0x991e8e20 | 0x400 | +| 206 | blk.9.attn_v.weight | 0x991e9220 | 0x113000 | +| 207 | blk.9.ffn_down.weight | 0x992fc220 | 0x330c00 | +| 208 | blk.9.ffn_down_exps.scale | 0x9962ce20 | 0x200 | +| 209 | blk.9.ffn_down_exps.weight | 0x9962d020 | 0x8820000 | +| 210 | blk.9.ffn_gate.weight | 0xa1e4d020 | 0x11b980 | +| 211 | blk.9.ffn_gate_inp.scale | 0xa1f689a0 | 0x2c00 | +| 212 | blk.9.ffn_gate_inp.weight | 0xa1f6b5a0 | 0x160000 | +| 213 | blk.9.ffn_gate_up_exps.weight | 0xa20cb5a0 | 0x5e88000 | +| 214 | blk.9.ffn_norm.weight | 0xa7f535a0 | 0x2c00 | +| 215 | blk.9.ffn_up.weight | 0xa7f561a0 | 0x11b980 | +| 216 | blk.9.layer_output_scale.weight | 0xa8071b20 | 0x4 | +| 217 | blk.9.post_attention_norm.weight | 0xa8071b40 | 0x2c00 | +| 218 | blk.9.post_ffw_norm.weight | 0xa8074740 | 0x2c00 | +| 219 | blk.9.post_ffw_norm_1.weight | 0xa8077340 | 0x2c00 | +| 220 | blk.9.post_ffw_norm_2.weight | 0xa8079f40 | 0x2c00 | +| 221 | blk.9.pre_ffw_norm_2.weight | 0xa807cb40 | 0x2c00 | +| 222 | blk.10.attn_k.weight | 0xa807f740 | 0x113000 | +| 223 | blk.10.attn_k_norm.weight | 0xa8192740 | 0x400 | +| 224 | blk.10.attn_norm.weight | 0xa8192b40 | 0x2c00 | +| 225 | blk.10.attn_output.weight | 0xa8195740 | 0x226000 | +| 226 | blk.10.attn_q.weight | 0xa83bb740 | 0x226000 | +| 227 | blk.10.attn_q_norm.weight | 0xa85e1740 | 0x400 | +| 228 | blk.10.attn_v.weight | 0xa85e1b40 | 0x113000 | +| 229 | blk.10.ffn_down.weight | 0xa86f4b40 | 0x330c00 | +| 230 | blk.10.ffn_down_exps.scale | 0xa8a25740 | 0x200 | +| 231 | blk.10.ffn_down_exps.weight | 0xa8a25940 | 0x8820000 | +| 232 | blk.10.ffn_gate.weight | 0xb1245940 | 0x11b980 | +| 233 | blk.10.ffn_gate_inp.scale | 0xb13612c0 | 0x2c00 | +| 234 | blk.10.ffn_gate_inp.weight | 0xb1363ec0 | 0x160000 | +| 235 | blk.10.ffn_gate_up_exps.weight | 0xb14c3ec0 | 0x5e88000 | +| 236 | blk.10.ffn_norm.weight | 0xb734bec0 | 0x2c00 | +| 237 | blk.10.ffn_up.weight | 0xb734eac0 | 0x11b980 | +| 238 | blk.10.layer_output_scale.weight | 0xb746a440 | 0x4 | +| 239 | blk.10.post_attention_norm.weight | 0xb746a460 | 0x2c00 | +| 240 | blk.10.post_ffw_norm.weight | 0xb746d060 | 0x2c00 | +| 241 | blk.10.post_ffw_norm_1.weight | 0xb746fc60 | 0x2c00 | +| 242 | blk.10.post_ffw_norm_2.weight | 0xb7472860 | 0x2c00 | +| 243 | blk.10.pre_ffw_norm_2.weight | 0xb7475460 | 0x2c00 | +| 244 | blk.11.attn_k.weight | 0xb7478060 | 0x89800 | +| 245 | blk.11.attn_k_norm.weight | 0xb7501860 | 0x800 | +| 246 | blk.11.attn_norm.weight | 0xb7502060 | 0x2c00 | +| 247 | blk.11.attn_output.weight | 0xb7504c60 | 0x44c000 | +| 248 | blk.11.attn_q.weight | 0xb7950c60 | 0x44c000 | +| 249 | blk.11.attn_q_norm.weight | 0xb7d9cc60 | 0x800 | +| 250 | blk.11.ffn_down.weight | 0xb7d9d460 | 0x330c00 | +| 251 | blk.11.ffn_down_exps.scale | 0xb80ce060 | 0x200 | +| 252 | blk.11.ffn_down_exps.weight | 0xb80ce260 | 0x8820000 | +| 253 | blk.11.ffn_gate.weight | 0xc08ee260 | 0x11b980 | +| 254 | blk.11.ffn_gate_inp.scale | 0xc0a09be0 | 0x2c00 | +| 255 | blk.11.ffn_gate_inp.weight | 0xc0a0c7e0 | 0x160000 | +| 256 | blk.11.ffn_gate_up_exps.weight | 0xc0b6c7e0 | 0x5e88000 | +| 257 | blk.11.ffn_norm.weight | 0xc69f47e0 | 0x2c00 | +| 258 | blk.11.ffn_up.weight | 0xc69f73e0 | 0x11b980 | +| 259 | blk.11.layer_output_scale.weight | 0xc6b12d60 | 0x4 | +| 260 | blk.11.post_attention_norm.weight | 0xc6b12d80 | 0x2c00 | +| 261 | blk.11.post_ffw_norm.weight | 0xc6b15980 | 0x2c00 | +| 262 | blk.11.post_ffw_norm_1.weight | 0xc6b18580 | 0x2c00 | +| 263 | blk.11.post_ffw_norm_2.weight | 0xc6b1b180 | 0x2c00 | +| 264 | blk.11.pre_ffw_norm_2.weight | 0xc6b1dd80 | 0x2c00 | +| 265 | blk.12.attn_k.weight | 0xc6b20980 | 0x113000 | +| 266 | blk.12.attn_k_norm.weight | 0xc6c33980 | 0x400 | +| 267 | blk.12.attn_norm.weight | 0xc6c33d80 | 0x2c00 | +| 268 | blk.12.attn_output.weight | 0xc6c36980 | 0x226000 | +| 269 | blk.12.attn_q.weight | 0xc6e5c980 | 0x226000 | +| 270 | blk.12.attn_q_norm.weight | 0xc7082980 | 0x400 | +| 271 | blk.12.attn_v.weight | 0xc7082d80 | 0x113000 | +| 272 | blk.12.ffn_down.weight | 0xc7195d80 | 0x330c00 | +| 273 | blk.12.ffn_down_exps.scale | 0xc74c6980 | 0x200 | +| 274 | blk.12.ffn_down_exps.weight | 0xc74c6b80 | 0x8820000 | +| 275 | blk.12.ffn_gate.weight | 0xcfce6b80 | 0x11b980 | +| 276 | blk.12.ffn_gate_inp.scale | 0xcfe02500 | 0x2c00 | +| 277 | blk.12.ffn_gate_inp.weight | 0xcfe05100 | 0x160000 | +| 278 | blk.12.ffn_gate_up_exps.weight | 0xcff65100 | 0x5e88000 | +| 279 | blk.12.ffn_norm.weight | 0xd5ded100 | 0x2c00 | +| 280 | blk.12.ffn_up.weight | 0xd5defd00 | 0x11b980 | +| 281 | blk.12.layer_output_scale.weight | 0xd5f0b680 | 0x4 | +| 282 | blk.12.post_attention_norm.weight | 0xd5f0b6a0 | 0x2c00 | +| 283 | blk.12.post_ffw_norm.weight | 0xd5f0e2a0 | 0x2c00 | +| 284 | blk.12.post_ffw_norm_1.weight | 0xd5f10ea0 | 0x2c00 | +| 285 | blk.12.post_ffw_norm_2.weight | 0xd5f13aa0 | 0x2c00 | +| 286 | blk.12.pre_ffw_norm_2.weight | 0xd5f166a0 | 0x2c00 | +| 287 | blk.13.attn_k.weight | 0xd5f192a0 | 0x113000 | +| 288 | blk.13.attn_k_norm.weight | 0xd602c2a0 | 0x400 | +| 289 | blk.13.attn_norm.weight | 0xd602c6a0 | 0x2c00 | +| 290 | blk.13.attn_output.weight | 0xd602f2a0 | 0x226000 | +| 291 | blk.13.attn_q.weight | 0xd62552a0 | 0x226000 | +| 292 | blk.13.attn_q_norm.weight | 0xd647b2a0 | 0x400 | +| 293 | blk.13.attn_v.weight | 0xd647b6a0 | 0x113000 | +| 294 | blk.13.ffn_down.weight | 0xd658e6a0 | 0x330c00 | +| 295 | blk.13.ffn_down_exps.scale | 0xd68bf2a0 | 0x200 | +| 296 | blk.13.ffn_down_exps.weight | 0xd68bf4a0 | 0x8820000 | +| 297 | blk.13.ffn_gate.weight | 0xdf0df4a0 | 0x11b980 | +| 298 | blk.13.ffn_gate_inp.scale | 0xdf1fae20 | 0x2c00 | +| 299 | blk.13.ffn_gate_inp.weight | 0xdf1fda20 | 0x160000 | +| 300 | blk.13.ffn_gate_up_exps.weight | 0xdf35da20 | 0x5e88000 | +| 301 | blk.13.ffn_norm.weight | 0xe51e5a20 | 0x2c00 | +| 302 | blk.13.ffn_up.weight | 0xe51e8620 | 0x11b980 | +| 303 | blk.13.layer_output_scale.weight | 0xe5303fa0 | 0x4 | +| 304 | blk.13.post_attention_norm.weight | 0xe5303fc0 | 0x2c00 | +| 305 | blk.13.post_ffw_norm.weight | 0xe5306bc0 | 0x2c00 | +| 306 | blk.13.post_ffw_norm_1.weight | 0xe53097c0 | 0x2c00 | +| 307 | blk.13.post_ffw_norm_2.weight | 0xe530c3c0 | 0x2c00 | +| 308 | blk.13.pre_ffw_norm_2.weight | 0xe530efc0 | 0x2c00 | +| 309 | blk.14.attn_k.weight | 0xe5311bc0 | 0x113000 | +| 310 | blk.14.attn_k_norm.weight | 0xe5424bc0 | 0x400 | +| 311 | blk.14.attn_norm.weight | 0xe5424fc0 | 0x2c00 | +| 312 | blk.14.attn_output.weight | 0xe5427bc0 | 0x226000 | +| 313 | blk.14.attn_q.weight | 0xe564dbc0 | 0x226000 | +| 314 | blk.14.attn_q_norm.weight | 0xe5873bc0 | 0x400 | +| 315 | blk.14.attn_v.weight | 0xe5873fc0 | 0x113000 | +| 316 | blk.14.ffn_down.weight | 0xe5986fc0 | 0x330c00 | +| 317 | blk.14.ffn_down_exps.scale | 0xe5cb7bc0 | 0x200 | +| 318 | blk.14.ffn_down_exps.weight | 0xe5cb7dc0 | 0x8820000 | +| 319 | blk.14.ffn_gate.weight | 0xee4d7dc0 | 0x11b980 | +| 320 | blk.14.ffn_gate_inp.scale | 0xee5f3740 | 0x2c00 | +| 321 | blk.14.ffn_gate_inp.weight | 0xee5f6340 | 0x160000 | +| 322 | blk.14.ffn_gate_up_exps.weight | 0xee756340 | 0x5e88000 | +| 323 | blk.14.ffn_norm.weight | 0xf45de340 | 0x2c00 | +| 324 | blk.14.ffn_up.weight | 0xf45e0f40 | 0x11b980 | +| 325 | blk.14.layer_output_scale.weight | 0xf46fc8c0 | 0x4 | +| 326 | blk.14.post_attention_norm.weight | 0xf46fc8e0 | 0x2c00 | +| 327 | blk.14.post_ffw_norm.weight | 0xf46ff4e0 | 0x2c00 | +| 328 | blk.14.post_ffw_norm_1.weight | 0xf47020e0 | 0x2c00 | +| 329 | blk.14.post_ffw_norm_2.weight | 0xf4704ce0 | 0x2c00 | +| 330 | blk.14.pre_ffw_norm_2.weight | 0xf47078e0 | 0x2c00 | +| 331 | blk.15.attn_k.weight | 0xf470a4e0 | 0x113000 | +| 332 | blk.15.attn_k_norm.weight | 0xf481d4e0 | 0x400 | +| 333 | blk.15.attn_norm.weight | 0xf481d8e0 | 0x2c00 | +| 334 | blk.15.attn_output.weight | 0xf48204e0 | 0x226000 | +| 335 | blk.15.attn_q.weight | 0xf4a464e0 | 0x226000 | +| 336 | blk.15.attn_q_norm.weight | 0xf4c6c4e0 | 0x400 | +| 337 | blk.15.attn_v.weight | 0xf4c6c8e0 | 0x113000 | +| 338 | blk.15.ffn_down.weight | 0xf4d7f8e0 | 0x330c00 | +| 339 | blk.15.ffn_down_exps.scale | 0xf50b04e0 | 0x200 | +| 340 | blk.15.ffn_down_exps.weight | 0xf50b06e0 | 0x8820000 | +| 341 | blk.15.ffn_gate.weight | 0xfd8d06e0 | 0x11b980 | +| 342 | blk.15.ffn_gate_inp.scale | 0xfd9ec060 | 0x2c00 | +| 343 | blk.15.ffn_gate_inp.weight | 0xfd9eec60 | 0x160000 | +| 344 | blk.15.ffn_gate_up_exps.weight | 0xfdb4ec60 | 0x5e88000 | +| 345 | blk.15.ffn_norm.weight | 0x1039d6c60 | 0x2c00 | +| 346 | blk.15.ffn_up.weight | 0x1039d9860 | 0x11b980 | +| 347 | blk.15.layer_output_scale.weight | 0x103af51e0 | 0x4 | +| 348 | blk.15.post_attention_norm.weight | 0x103af5200 | 0x2c00 | +| 349 | blk.15.post_ffw_norm.weight | 0x103af7e00 | 0x2c00 | +| 350 | blk.15.post_ffw_norm_1.weight | 0x103afaa00 | 0x2c00 | +| 351 | blk.15.post_ffw_norm_2.weight | 0x103afd600 | 0x2c00 | +| 352 | blk.15.pre_ffw_norm_2.weight | 0x103b00200 | 0x2c00 | +| 353 | blk.16.attn_k.weight | 0x103b02e00 | 0x113000 | +| 354 | blk.16.attn_k_norm.weight | 0x103c15e00 | 0x400 | +| 355 | blk.16.attn_norm.weight | 0x103c16200 | 0x2c00 | +| 356 | blk.16.attn_output.weight | 0x103c18e00 | 0x226000 | +| 357 | blk.16.attn_q.weight | 0x103e3ee00 | 0x226000 | +| 358 | blk.16.attn_q_norm.weight | 0x104064e00 | 0x400 | +| 359 | blk.16.attn_v.weight | 0x104065200 | 0x113000 | +| 360 | blk.16.ffn_down.weight | 0x104178200 | 0x330c00 | +| 361 | blk.16.ffn_down_exps.scale | 0x1044a8e00 | 0x200 | +| 362 | blk.16.ffn_down_exps.weight | 0x1044a9000 | 0x8820000 | +| 363 | blk.16.ffn_gate.weight | 0x10ccc9000 | 0x11b980 | +| 364 | blk.16.ffn_gate_inp.scale | 0x10cde4980 | 0x2c00 | +| 365 | blk.16.ffn_gate_inp.weight | 0x10cde7580 | 0x160000 | +| 366 | blk.16.ffn_gate_up_exps.weight | 0x10cf47580 | 0x5e88000 | +| 367 | blk.16.ffn_norm.weight | 0x112dcf580 | 0x2c00 | +| 368 | blk.16.ffn_up.weight | 0x112dd2180 | 0x11b980 | +| 369 | blk.16.layer_output_scale.weight | 0x112eedb00 | 0x4 | +| 370 | blk.16.post_attention_norm.weight | 0x112eedb20 | 0x2c00 | +| 371 | blk.16.post_ffw_norm.weight | 0x112ef0720 | 0x2c00 | +| 372 | blk.16.post_ffw_norm_1.weight | 0x112ef3320 | 0x2c00 | +| 373 | blk.16.post_ffw_norm_2.weight | 0x112ef5f20 | 0x2c00 | +| 374 | blk.16.pre_ffw_norm_2.weight | 0x112ef8b20 | 0x2c00 | +| 375 | blk.17.attn_k.weight | 0x112efb720 | 0x89800 | +| 376 | blk.17.attn_k_norm.weight | 0x112f84f20 | 0x800 | +| 377 | blk.17.attn_norm.weight | 0x112f85720 | 0x2c00 | +| 378 | blk.17.attn_output.weight | 0x112f88320 | 0x44c000 | +| 379 | blk.17.attn_q.weight | 0x1133d4320 | 0x44c000 | +| 380 | blk.17.attn_q_norm.weight | 0x113820320 | 0x800 | +| 381 | blk.17.ffn_down.weight | 0x113820b20 | 0x330c00 | +| 382 | blk.17.ffn_down_exps.scale | 0x113b51720 | 0x200 | +| 383 | blk.17.ffn_down_exps.weight | 0x113b51920 | 0x8820000 | +| 384 | blk.17.ffn_gate.weight | 0x11c371920 | 0x11b980 | +| 385 | blk.17.ffn_gate_inp.scale | 0x11c48d2a0 | 0x2c00 | +| 386 | blk.17.ffn_gate_inp.weight | 0x11c48fea0 | 0x160000 | +| 387 | blk.17.ffn_gate_up_exps.weight | 0x11c5efea0 | 0x5e88000 | +| 388 | blk.17.ffn_norm.weight | 0x122477ea0 | 0x2c00 | +| 389 | blk.17.ffn_up.weight | 0x12247aaa0 | 0x11b980 | +| 390 | blk.17.layer_output_scale.weight | 0x122596420 | 0x4 | +| 391 | blk.17.post_attention_norm.weight | 0x122596440 | 0x2c00 | +| 392 | blk.17.post_ffw_norm.weight | 0x122599040 | 0x2c00 | +| 393 | blk.17.post_ffw_norm_1.weight | 0x12259bc40 | 0x2c00 | +| 394 | blk.17.post_ffw_norm_2.weight | 0x12259e840 | 0x2c00 | +| 395 | blk.17.pre_ffw_norm_2.weight | 0x1225a1440 | 0x2c00 | +| 396 | blk.18.attn_k.weight | 0x1225a4040 | 0x113000 | +| 397 | blk.18.attn_k_norm.weight | 0x1226b7040 | 0x400 | +| 398 | blk.18.attn_norm.weight | 0x1226b7440 | 0x2c00 | +| 399 | blk.18.attn_output.weight | 0x1226ba040 | 0x226000 | +| 400 | blk.18.attn_q.weight | 0x1228e0040 | 0x226000 | +| 401 | blk.18.attn_q_norm.weight | 0x122b06040 | 0x400 | +| 402 | blk.18.attn_v.weight | 0x122b06440 | 0x113000 | +| 403 | blk.18.ffn_down.weight | 0x122c19440 | 0x330c00 | +| 404 | blk.18.ffn_down_exps.scale | 0x122f4a040 | 0x200 | +| 405 | blk.18.ffn_down_exps.weight | 0x122f4a240 | 0x8820000 | +| 406 | blk.18.ffn_gate.weight | 0x12b76a240 | 0x11b980 | +| 407 | blk.18.ffn_gate_inp.scale | 0x12b885bc0 | 0x2c00 | +| 408 | blk.18.ffn_gate_inp.weight | 0x12b8887c0 | 0x160000 | +| 409 | blk.18.ffn_gate_up_exps.weight | 0x12b9e87c0 | 0x5e88000 | +| 410 | blk.18.ffn_norm.weight | 0x1318707c0 | 0x2c00 | +| 411 | blk.18.ffn_up.weight | 0x1318733c0 | 0x11b980 | +| 412 | blk.18.layer_output_scale.weight | 0x13198ed40 | 0x4 | +| 413 | blk.18.post_attention_norm.weight | 0x13198ed60 | 0x2c00 | +| 414 | blk.18.post_ffw_norm.weight | 0x131991960 | 0x2c00 | +| 415 | blk.18.post_ffw_norm_1.weight | 0x131994560 | 0x2c00 | +| 416 | blk.18.post_ffw_norm_2.weight | 0x131997160 | 0x2c00 | +| 417 | blk.18.pre_ffw_norm_2.weight | 0x131999d60 | 0x2c00 | +| 418 | blk.19.attn_k.weight | 0x13199c960 | 0x113000 | +| 419 | blk.19.attn_k_norm.weight | 0x131aaf960 | 0x400 | +| 420 | blk.19.attn_norm.weight | 0x131aafd60 | 0x2c00 | +| 421 | blk.19.attn_output.weight | 0x131ab2960 | 0x226000 | +| 422 | blk.19.attn_q.weight | 0x131cd8960 | 0x226000 | +| 423 | blk.19.attn_q_norm.weight | 0x131efe960 | 0x400 | +| 424 | blk.19.attn_v.weight | 0x131efed60 | 0x113000 | +| 425 | blk.19.ffn_down.weight | 0x132011d60 | 0x330c00 | +| 426 | blk.19.ffn_down_exps.scale | 0x132342960 | 0x200 | +| 427 | blk.19.ffn_down_exps.weight | 0x132342b60 | 0x8820000 | +| 428 | blk.19.ffn_gate.weight | 0x13ab62b60 | 0x11b980 | +| 429 | blk.19.ffn_gate_inp.scale | 0x13ac7e4e0 | 0x2c00 | +| 430 | blk.19.ffn_gate_inp.weight | 0x13ac810e0 | 0x160000 | +| 431 | blk.19.ffn_gate_up_exps.weight | 0x13ade10e0 | 0x5e88000 | +| 432 | blk.19.ffn_norm.weight | 0x140c690e0 | 0x2c00 | +| 433 | blk.19.ffn_up.weight | 0x140c6bce0 | 0x11b980 | +| 434 | blk.19.layer_output_scale.weight | 0x140d87660 | 0x4 | +| 435 | blk.19.post_attention_norm.weight | 0x140d87680 | 0x2c00 | +| 436 | blk.19.post_ffw_norm.weight | 0x140d8a280 | 0x2c00 | +| 437 | blk.19.post_ffw_norm_1.weight | 0x140d8ce80 | 0x2c00 | +| 438 | blk.19.post_ffw_norm_2.weight | 0x140d8fa80 | 0x2c00 | +| 439 | blk.19.pre_ffw_norm_2.weight | 0x140d92680 | 0x2c00 | +| 440 | blk.20.attn_k.weight | 0x140d95280 | 0x113000 | +| 441 | blk.20.attn_k_norm.weight | 0x140ea8280 | 0x400 | +| 442 | blk.20.attn_norm.weight | 0x140ea8680 | 0x2c00 | +| 443 | blk.20.attn_output.weight | 0x140eab280 | 0x226000 | +| 444 | blk.20.attn_q.weight | 0x1410d1280 | 0x226000 | +| 445 | blk.20.attn_q_norm.weight | 0x1412f7280 | 0x400 | +| 446 | blk.20.attn_v.weight | 0x1412f7680 | 0x113000 | +| 447 | blk.20.ffn_down.weight | 0x14140a680 | 0x330c00 | +| 448 | blk.20.ffn_down_exps.scale | 0x14173b280 | 0x200 | +| 449 | blk.20.ffn_down_exps.weight | 0x14173b480 | 0x8820000 | +| 450 | blk.20.ffn_gate.weight | 0x149f5b480 | 0x11b980 | +| 451 | blk.20.ffn_gate_inp.scale | 0x14a076e00 | 0x2c00 | +| 452 | blk.20.ffn_gate_inp.weight | 0x14a079a00 | 0x160000 | +| 453 | blk.20.ffn_gate_up_exps.weight | 0x14a1d9a00 | 0x5e88000 | +| 454 | blk.20.ffn_norm.weight | 0x150061a00 | 0x2c00 | +| 455 | blk.20.ffn_up.weight | 0x150064600 | 0x11b980 | +| 456 | blk.20.layer_output_scale.weight | 0x15017ff80 | 0x4 | +| 457 | blk.20.post_attention_norm.weight | 0x15017ffa0 | 0x2c00 | +| 458 | blk.20.post_ffw_norm.weight | 0x150182ba0 | 0x2c00 | +| 459 | blk.20.post_ffw_norm_1.weight | 0x1501857a0 | 0x2c00 | +| 460 | blk.20.post_ffw_norm_2.weight | 0x1501883a0 | 0x2c00 | +| 461 | blk.20.pre_ffw_norm_2.weight | 0x15018afa0 | 0x2c00 | +| 462 | blk.21.attn_k.weight | 0x15018dba0 | 0x113000 | +| 463 | blk.21.attn_k_norm.weight | 0x1502a0ba0 | 0x400 | +| 464 | blk.21.attn_norm.weight | 0x1502a0fa0 | 0x2c00 | +| 465 | blk.21.attn_output.weight | 0x1502a3ba0 | 0x226000 | +| 466 | blk.21.attn_q.weight | 0x1504c9ba0 | 0x226000 | +| 467 | blk.21.attn_q_norm.weight | 0x1506efba0 | 0x400 | +| 468 | blk.21.attn_v.weight | 0x1506effa0 | 0x113000 | +| 469 | blk.21.ffn_down.weight | 0x150802fa0 | 0x330c00 | +| 470 | blk.21.ffn_down_exps.scale | 0x150b33ba0 | 0x200 | +| 471 | blk.21.ffn_down_exps.weight | 0x150b33da0 | 0x8820000 | +| 472 | blk.21.ffn_gate.weight | 0x159353da0 | 0x11b980 | +| 473 | blk.21.ffn_gate_inp.scale | 0x15946f720 | 0x2c00 | +| 474 | blk.21.ffn_gate_inp.weight | 0x159472320 | 0x160000 | +| 475 | blk.21.ffn_gate_up_exps.weight | 0x1595d2320 | 0x5e88000 | +| 476 | blk.21.ffn_norm.weight | 0x15f45a320 | 0x2c00 | +| 477 | blk.21.ffn_up.weight | 0x15f45cf20 | 0x11b980 | +| 478 | blk.21.layer_output_scale.weight | 0x15f5788a0 | 0x4 | +| 479 | blk.21.post_attention_norm.weight | 0x15f5788c0 | 0x2c00 | +| 480 | blk.21.post_ffw_norm.weight | 0x15f57b4c0 | 0x2c00 | +| 481 | blk.21.post_ffw_norm_1.weight | 0x15f57e0c0 | 0x2c00 | +| 482 | blk.21.post_ffw_norm_2.weight | 0x15f580cc0 | 0x2c00 | +| 483 | blk.21.pre_ffw_norm_2.weight | 0x15f5838c0 | 0x2c00 | +| 484 | blk.22.attn_k.weight | 0x15f5864c0 | 0x113000 | +| 485 | blk.22.attn_k_norm.weight | 0x15f6994c0 | 0x400 | +| 486 | blk.22.attn_norm.weight | 0x15f6998c0 | 0x2c00 | +| 487 | blk.22.attn_output.weight | 0x15f69c4c0 | 0x226000 | +| 488 | blk.22.attn_q.weight | 0x15f8c24c0 | 0x226000 | +| 489 | blk.22.attn_q_norm.weight | 0x15fae84c0 | 0x400 | +| 490 | blk.22.attn_v.weight | 0x15fae88c0 | 0x113000 | +| 491 | blk.22.ffn_down.weight | 0x15fbfb8c0 | 0x330c00 | +| 492 | blk.22.ffn_down_exps.scale | 0x15ff2c4c0 | 0x200 | +| 493 | blk.22.ffn_down_exps.weight | 0x15ff2c6c0 | 0x8820000 | +| 494 | blk.22.ffn_gate.weight | 0x16874c6c0 | 0x11b980 | +| 495 | blk.22.ffn_gate_inp.scale | 0x168868040 | 0x2c00 | +| 496 | blk.22.ffn_gate_inp.weight | 0x16886ac40 | 0x160000 | +| 497 | blk.22.ffn_gate_up_exps.weight | 0x1689cac40 | 0x5e88000 | +| 498 | blk.22.ffn_norm.weight | 0x16e852c40 | 0x2c00 | +| 499 | blk.22.ffn_up.weight | 0x16e855840 | 0x11b980 | +| 500 | blk.22.layer_output_scale.weight | 0x16e9711c0 | 0x4 | +| 501 | blk.22.post_attention_norm.weight | 0x16e9711e0 | 0x2c00 | +| 502 | blk.22.post_ffw_norm.weight | 0x16e973de0 | 0x2c00 | +| 503 | blk.22.post_ffw_norm_1.weight | 0x16e9769e0 | 0x2c00 | +| 504 | blk.22.post_ffw_norm_2.weight | 0x16e9795e0 | 0x2c00 | +| 505 | blk.22.pre_ffw_norm_2.weight | 0x16e97c1e0 | 0x2c00 | +| 506 | blk.23.attn_k.weight | 0x16e97ede0 | 0x89800 | +| 507 | blk.23.attn_k_norm.weight | 0x16ea085e0 | 0x800 | +| 508 | blk.23.attn_norm.weight | 0x16ea08de0 | 0x2c00 | +| 509 | blk.23.attn_output.weight | 0x16ea0b9e0 | 0x44c000 | +| 510 | blk.23.attn_q.weight | 0x16ee579e0 | 0x44c000 | +| 511 | blk.23.attn_q_norm.weight | 0x16f2a39e0 | 0x800 | +| 512 | blk.23.ffn_down.weight | 0x16f2a41e0 | 0x330c00 | +| 513 | blk.23.ffn_down_exps.scale | 0x16f5d4de0 | 0x200 | +| 514 | blk.23.ffn_down_exps.weight | 0x16f5d4fe0 | 0x8820000 | +| 515 | blk.23.ffn_gate.weight | 0x177df4fe0 | 0x11b980 | +| 516 | blk.23.ffn_gate_inp.scale | 0x177f10960 | 0x2c00 | +| 517 | blk.23.ffn_gate_inp.weight | 0x177f13560 | 0x160000 | +| 518 | blk.23.ffn_gate_up_exps.weight | 0x178073560 | 0x5e88000 | +| 519 | blk.23.ffn_norm.weight | 0x17defb560 | 0x2c00 | +| 520 | blk.23.ffn_up.weight | 0x17defe160 | 0x11b980 | +| 521 | blk.23.layer_output_scale.weight | 0x17e019ae0 | 0x4 | +| 522 | blk.23.post_attention_norm.weight | 0x17e019b00 | 0x2c00 | +| 523 | blk.23.post_ffw_norm.weight | 0x17e01c700 | 0x2c00 | +| 524 | blk.23.post_ffw_norm_1.weight | 0x17e01f300 | 0x2c00 | +| 525 | blk.23.post_ffw_norm_2.weight | 0x17e021f00 | 0x2c00 | +| 526 | blk.23.pre_ffw_norm_2.weight | 0x17e024b00 | 0x2c00 | +| 527 | blk.24.attn_k.weight | 0x17e027700 | 0x113000 | +| 528 | blk.24.attn_k_norm.weight | 0x17e13a700 | 0x400 | +| 529 | blk.24.attn_norm.weight | 0x17e13ab00 | 0x2c00 | +| 530 | blk.24.attn_output.weight | 0x17e13d700 | 0x226000 | +| 531 | blk.24.attn_q.weight | 0x17e363700 | 0x226000 | +| 532 | blk.24.attn_q_norm.weight | 0x17e589700 | 0x400 | +| 533 | blk.24.attn_v.weight | 0x17e589b00 | 0x113000 | +| 534 | blk.24.ffn_down.weight | 0x17e69cb00 | 0x330c00 | +| 535 | blk.24.ffn_down_exps.scale | 0x17e9cd700 | 0x200 | +| 536 | blk.24.ffn_down_exps.weight | 0x17e9cd900 | 0x8820000 | +| 537 | blk.24.ffn_gate.weight | 0x1871ed900 | 0x11b980 | +| 538 | blk.24.ffn_gate_inp.scale | 0x187309280 | 0x2c00 | +| 539 | blk.24.ffn_gate_inp.weight | 0x18730be80 | 0x160000 | +| 540 | blk.24.ffn_gate_up_exps.weight | 0x18746be80 | 0x5e88000 | +| 541 | blk.24.ffn_norm.weight | 0x18d2f3e80 | 0x2c00 | +| 542 | blk.24.ffn_up.weight | 0x18d2f6a80 | 0x11b980 | +| 543 | blk.24.layer_output_scale.weight | 0x18d412400 | 0x4 | +| 544 | blk.24.post_attention_norm.weight | 0x18d412420 | 0x2c00 | +| 545 | blk.24.post_ffw_norm.weight | 0x18d415020 | 0x2c00 | +| 546 | blk.24.post_ffw_norm_1.weight | 0x18d417c20 | 0x2c00 | +| 547 | blk.24.post_ffw_norm_2.weight | 0x18d41a820 | 0x2c00 | +| 548 | blk.24.pre_ffw_norm_2.weight | 0x18d41d420 | 0x2c00 | +| 549 | blk.25.attn_k.weight | 0x18d420020 | 0x113000 | +| 550 | blk.25.attn_k_norm.weight | 0x18d533020 | 0x400 | +| 551 | blk.25.attn_norm.weight | 0x18d533420 | 0x2c00 | +| 552 | blk.25.attn_output.weight | 0x18d536020 | 0x226000 | +| 553 | blk.25.attn_q.weight | 0x18d75c020 | 0x226000 | +| 554 | blk.25.attn_q_norm.weight | 0x18d982020 | 0x400 | +| 555 | blk.25.attn_v.weight | 0x18d982420 | 0x113000 | +| 556 | blk.25.ffn_down.weight | 0x18da95420 | 0x330c00 | +| 557 | blk.25.ffn_down_exps.scale | 0x18ddc6020 | 0x200 | +| 558 | blk.25.ffn_down_exps.weight | 0x18ddc6220 | 0x8820000 | +| 559 | blk.25.ffn_gate.weight | 0x1965e6220 | 0x11b980 | +| 560 | blk.25.ffn_gate_inp.scale | 0x196701ba0 | 0x2c00 | +| 561 | blk.25.ffn_gate_inp.weight | 0x1967047a0 | 0x160000 | +| 562 | blk.25.ffn_gate_up_exps.weight | 0x1968647a0 | 0x5e88000 | +| 563 | blk.25.ffn_norm.weight | 0x19c6ec7a0 | 0x2c00 | +| 564 | blk.25.ffn_up.weight | 0x19c6ef3a0 | 0x11b980 | +| 565 | blk.25.layer_output_scale.weight | 0x19c80ad20 | 0x4 | +| 566 | blk.25.post_attention_norm.weight | 0x19c80ad40 | 0x2c00 | +| 567 | blk.25.post_ffw_norm.weight | 0x19c80d940 | 0x2c00 | +| 568 | blk.25.post_ffw_norm_1.weight | 0x19c810540 | 0x2c00 | +| 569 | blk.25.post_ffw_norm_2.weight | 0x19c813140 | 0x2c00 | +| 570 | blk.25.pre_ffw_norm_2.weight | 0x19c815d40 | 0x2c00 | +| 571 | blk.26.attn_k.weight | 0x19c818940 | 0x113000 | +| 572 | blk.26.attn_k_norm.weight | 0x19c92b940 | 0x400 | +| 573 | blk.26.attn_norm.weight | 0x19c92bd40 | 0x2c00 | +| 574 | blk.26.attn_output.weight | 0x19c92e940 | 0x226000 | +| 575 | blk.26.attn_q.weight | 0x19cb54940 | 0x226000 | +| 576 | blk.26.attn_q_norm.weight | 0x19cd7a940 | 0x400 | +| 577 | blk.26.attn_v.weight | 0x19cd7ad40 | 0x113000 | +| 578 | blk.26.ffn_down.weight | 0x19ce8dd40 | 0x330c00 | +| 579 | blk.26.ffn_down_exps.scale | 0x19d1be940 | 0x200 | +| 580 | blk.26.ffn_down_exps.weight | 0x19d1beb40 | 0x8820000 | +| 581 | blk.26.ffn_gate.weight | 0x1a59deb40 | 0x11b980 | +| 582 | blk.26.ffn_gate_inp.scale | 0x1a5afa4c0 | 0x2c00 | +| 583 | blk.26.ffn_gate_inp.weight | 0x1a5afd0c0 | 0x160000 | +| 584 | blk.26.ffn_gate_up_exps.weight | 0x1a5c5d0c0 | 0x5e88000 | +| 585 | blk.26.ffn_norm.weight | 0x1abae50c0 | 0x2c00 | +| 586 | blk.26.ffn_up.weight | 0x1abae7cc0 | 0x11b980 | +| 587 | blk.26.layer_output_scale.weight | 0x1abc03640 | 0x4 | +| 588 | blk.26.post_attention_norm.weight | 0x1abc03660 | 0x2c00 | +| 589 | blk.26.post_ffw_norm.weight | 0x1abc06260 | 0x2c00 | +| 590 | blk.26.post_ffw_norm_1.weight | 0x1abc08e60 | 0x2c00 | +| 591 | blk.26.post_ffw_norm_2.weight | 0x1abc0ba60 | 0x2c00 | +| 592 | blk.26.pre_ffw_norm_2.weight | 0x1abc0e660 | 0x2c00 | +| 593 | blk.27.attn_k.weight | 0x1abc11260 | 0x113000 | +| 594 | blk.27.attn_k_norm.weight | 0x1abd24260 | 0x400 | +| 595 | blk.27.attn_norm.weight | 0x1abd24660 | 0x2c00 | +| 596 | blk.27.attn_output.weight | 0x1abd27260 | 0x226000 | +| 597 | blk.27.attn_q.weight | 0x1abf4d260 | 0x226000 | +| 598 | blk.27.attn_q_norm.weight | 0x1ac173260 | 0x400 | +| 599 | blk.27.attn_v.weight | 0x1ac173660 | 0x113000 | +| 600 | blk.27.ffn_down.weight | 0x1ac286660 | 0x330c00 | +| 601 | blk.27.ffn_down_exps.scale | 0x1ac5b7260 | 0x200 | +| 602 | blk.27.ffn_down_exps.weight | 0x1ac5b7460 | 0x8820000 | +| 603 | blk.27.ffn_gate.weight | 0x1b4dd7460 | 0x11b980 | +| 604 | blk.27.ffn_gate_inp.scale | 0x1b4ef2de0 | 0x2c00 | +| 605 | blk.27.ffn_gate_inp.weight | 0x1b4ef59e0 | 0x160000 | +| 606 | blk.27.ffn_gate_up_exps.weight | 0x1b50559e0 | 0x5e88000 | +| 607 | blk.27.ffn_norm.weight | 0x1baedd9e0 | 0x2c00 | +| 608 | blk.27.ffn_up.weight | 0x1baee05e0 | 0x11b980 | +| 609 | blk.27.layer_output_scale.weight | 0x1baffbf60 | 0x4 | +| 610 | blk.27.post_attention_norm.weight | 0x1baffbf80 | 0x2c00 | +| 611 | blk.27.post_ffw_norm.weight | 0x1baffeb80 | 0x2c00 | +| 612 | blk.27.post_ffw_norm_1.weight | 0x1bb001780 | 0x2c00 | +| 613 | blk.27.post_ffw_norm_2.weight | 0x1bb004380 | 0x2c00 | +| 614 | blk.27.pre_ffw_norm_2.weight | 0x1bb006f80 | 0x2c00 | +| 615 | blk.28.attn_k.weight | 0x1bb009b80 | 0x113000 | +| 616 | blk.28.attn_k_norm.weight | 0x1bb11cb80 | 0x400 | +| 617 | blk.28.attn_norm.weight | 0x1bb11cf80 | 0x2c00 | +| 618 | blk.28.attn_output.weight | 0x1bb11fb80 | 0x226000 | +| 619 | blk.28.attn_q.weight | 0x1bb345b80 | 0x226000 | +| 620 | blk.28.attn_q_norm.weight | 0x1bb56bb80 | 0x400 | +| 621 | blk.28.attn_v.weight | 0x1bb56bf80 | 0x113000 | +| 622 | blk.28.ffn_down.weight | 0x1bb67ef80 | 0x330c00 | +| 623 | blk.28.ffn_down_exps.scale | 0x1bb9afb80 | 0x200 | +| 624 | blk.28.ffn_down_exps.weight | 0x1bb9afd80 | 0x8820000 | +| 625 | blk.28.ffn_gate.weight | 0x1c41cfd80 | 0x11b980 | +| 626 | blk.28.ffn_gate_inp.scale | 0x1c42eb700 | 0x2c00 | +| 627 | blk.28.ffn_gate_inp.weight | 0x1c42ee300 | 0x160000 | +| 628 | blk.28.ffn_gate_up_exps.weight | 0x1c444e300 | 0x5e88000 | +| 629 | blk.28.ffn_norm.weight | 0x1ca2d6300 | 0x2c00 | +| 630 | blk.28.ffn_up.weight | 0x1ca2d8f00 | 0x11b980 | +| 631 | blk.28.layer_output_scale.weight | 0x1ca3f4880 | 0x4 | +| 632 | blk.28.post_attention_norm.weight | 0x1ca3f48a0 | 0x2c00 | +| 633 | blk.28.post_ffw_norm.weight | 0x1ca3f74a0 | 0x2c00 | +| 634 | blk.28.post_ffw_norm_1.weight | 0x1ca3fa0a0 | 0x2c00 | +| 635 | blk.28.post_ffw_norm_2.weight | 0x1ca3fcca0 | 0x2c00 | +| 636 | blk.28.pre_ffw_norm_2.weight | 0x1ca3ff8a0 | 0x2c00 | +| 637 | blk.29.attn_k.weight | 0x1ca4024a0 | 0x89800 | +| 638 | blk.29.attn_k_norm.weight | 0x1ca48bca0 | 0x800 | +| 639 | blk.29.attn_norm.weight | 0x1ca48c4a0 | 0x2c00 | +| 640 | blk.29.attn_output.weight | 0x1ca48f0a0 | 0x44c000 | +| 641 | blk.29.attn_q.weight | 0x1ca8db0a0 | 0x44c000 | +| 642 | blk.29.attn_q_norm.weight | 0x1cad270a0 | 0x800 | +| 643 | blk.29.ffn_down.weight | 0x1cad278a0 | 0x330c00 | +| 644 | blk.29.ffn_down_exps.scale | 0x1cb0584a0 | 0x200 | +| 645 | blk.29.ffn_down_exps.weight | 0x1cb0586a0 | 0x8820000 | +| 646 | blk.29.ffn_gate.weight | 0x1d38786a0 | 0x11b980 | +| 647 | blk.29.ffn_gate_inp.scale | 0x1d3994020 | 0x2c00 | +| 648 | blk.29.ffn_gate_inp.weight | 0x1d3996c20 | 0x160000 | +| 649 | blk.29.ffn_gate_up_exps.weight | 0x1d3af6c20 | 0x5e88000 | +| 650 | blk.29.ffn_norm.weight | 0x1d997ec20 | 0x2c00 | +| 651 | blk.29.ffn_up.weight | 0x1d9981820 | 0x11b980 | +| 652 | blk.29.layer_output_scale.weight | 0x1d9a9d1a0 | 0x4 | +| 653 | blk.29.post_attention_norm.weight | 0x1d9a9d1c0 | 0x2c00 | +| 654 | blk.29.post_ffw_norm.weight | 0x1d9a9fdc0 | 0x2c00 | +| 655 | blk.29.post_ffw_norm_1.weight | 0x1d9aa29c0 | 0x2c00 | +| 656 | blk.29.post_ffw_norm_2.weight | 0x1d9aa55c0 | 0x2c00 | +| 657 | blk.29.pre_ffw_norm_2.weight | 0x1d9aa81c0 | 0x2c00 | + +### Base Tensor Group : ~738M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :----------------- | :------------------------------- | :---------------- | :-------------------- | :--- | ------: | +| 0 | output_norm.weight | Output Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 1 | rope_freqs.weight | Rope_Freqs (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 2 | token_embd.weight | Token Embedding (W) | (~738M) 738197504 | 2816 x 262144 x 1 x 1 | Q2_K | 2.6250 | + +- Total elements in base: (~738M) 738200576 +- Percentage of total elements: 2.93% +- Bits per Weight (BPW) for base: 2.6251 bits + + +### Block 0 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 3 | blk.0.attn_k.weight | Block 0 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 4 | blk.0.attn_k_norm.weight | Block 0 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 5 | blk.0.attn_norm.weight | Block 0 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 6 | blk.0.attn_output.weight | Block 0 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 7 | blk.0.attn_q.weight | Block 0 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 8 | blk.0.attn_q_norm.weight | Block 0 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 9 | blk.0.attn_v.weight | Block 0 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 10 | blk.0.ffn_down.weight | Block 0 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 11 | blk.0.ffn_down_exps.scale | Block 0 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 12 | blk.0.ffn_down_exps.weight | Block 0 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 13 | blk.0.ffn_gate.weight | Block 0 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 14 | blk.0.ffn_gate_inp.scale | Block 0 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 15 | blk.0.ffn_gate_inp.weight | Block 0 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 16 | blk.0.ffn_gate_up_exps.weight | Block 0 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 17 | blk.0.ffn_norm.weight | Block 0 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 18 | blk.0.ffn_up.weight | Block 0 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 19 | blk.0.layer_output_scale.weight | Block 0 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 20 | blk.0.post_attention_norm.weight | Block 0 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 21 | blk.0.post_ffw_norm.weight | Block 0 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 22 | blk.0.post_ffw_norm_1.weight | Block 0 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 23 | blk.0.post_ffw_norm_2.weight | Block 0 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 24 | blk.0.pre_ffw_norm_2.weight | Block 0 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.0: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.0: 2.5139 bits + + +### Block 1 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 25 | blk.1.attn_k.weight | Block 1 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 26 | blk.1.attn_k_norm.weight | Block 1 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 27 | blk.1.attn_norm.weight | Block 1 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 28 | blk.1.attn_output.weight | Block 1 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 29 | blk.1.attn_q.weight | Block 1 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 30 | blk.1.attn_q_norm.weight | Block 1 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 31 | blk.1.attn_v.weight | Block 1 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 32 | blk.1.ffn_down.weight | Block 1 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 33 | blk.1.ffn_down_exps.scale | Block 1 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 34 | blk.1.ffn_down_exps.weight | Block 1 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 35 | blk.1.ffn_gate.weight | Block 1 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 36 | blk.1.ffn_gate_inp.scale | Block 1 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 37 | blk.1.ffn_gate_inp.weight | Block 1 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 38 | blk.1.ffn_gate_up_exps.weight | Block 1 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 39 | blk.1.ffn_norm.weight | Block 1 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 40 | blk.1.ffn_up.weight | Block 1 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 41 | blk.1.layer_output_scale.weight | Block 1 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 42 | blk.1.post_attention_norm.weight | Block 1 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 43 | blk.1.post_ffw_norm.weight | Block 1 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 44 | blk.1.post_ffw_norm_1.weight | Block 1 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 45 | blk.1.post_ffw_norm_2.weight | Block 1 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 46 | blk.1.pre_ffw_norm_2.weight | Block 1 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.1: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.1: 2.5139 bits + + +### Block 2 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 47 | blk.2.attn_k.weight | Block 2 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 48 | blk.2.attn_k_norm.weight | Block 2 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 49 | blk.2.attn_norm.weight | Block 2 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 50 | blk.2.attn_output.weight | Block 2 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 51 | blk.2.attn_q.weight | Block 2 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 52 | blk.2.attn_q_norm.weight | Block 2 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 53 | blk.2.attn_v.weight | Block 2 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 54 | blk.2.ffn_down.weight | Block 2 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 55 | blk.2.ffn_down_exps.scale | Block 2 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 56 | blk.2.ffn_down_exps.weight | Block 2 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 57 | blk.2.ffn_gate.weight | Block 2 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 58 | blk.2.ffn_gate_inp.scale | Block 2 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 59 | blk.2.ffn_gate_inp.weight | Block 2 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 60 | blk.2.ffn_gate_up_exps.weight | Block 2 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 61 | blk.2.ffn_norm.weight | Block 2 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 62 | blk.2.ffn_up.weight | Block 2 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 63 | blk.2.layer_output_scale.weight | Block 2 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 64 | blk.2.post_attention_norm.weight | Block 2 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 65 | blk.2.post_ffw_norm.weight | Block 2 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 66 | blk.2.post_ffw_norm_1.weight | Block 2 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 67 | blk.2.post_ffw_norm_2.weight | Block 2 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 68 | blk.2.pre_ffw_norm_2.weight | Block 2 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.2: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.2: 2.5139 bits + + +### Block 3 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 69 | blk.3.attn_k.weight | Block 3 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 70 | blk.3.attn_k_norm.weight | Block 3 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 71 | blk.3.attn_norm.weight | Block 3 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 72 | blk.3.attn_output.weight | Block 3 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 73 | blk.3.attn_q.weight | Block 3 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 74 | blk.3.attn_q_norm.weight | Block 3 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 75 | blk.3.attn_v.weight | Block 3 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 76 | blk.3.ffn_down.weight | Block 3 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 77 | blk.3.ffn_down_exps.scale | Block 3 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 78 | blk.3.ffn_down_exps.weight | Block 3 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 79 | blk.3.ffn_gate.weight | Block 3 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 80 | blk.3.ffn_gate_inp.scale | Block 3 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 81 | blk.3.ffn_gate_inp.weight | Block 3 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 82 | blk.3.ffn_gate_up_exps.weight | Block 3 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 83 | blk.3.ffn_norm.weight | Block 3 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 84 | blk.3.ffn_up.weight | Block 3 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 85 | blk.3.layer_output_scale.weight | Block 3 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 86 | blk.3.post_attention_norm.weight | Block 3 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 87 | blk.3.post_ffw_norm.weight | Block 3 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 88 | blk.3.post_ffw_norm_1.weight | Block 3 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 89 | blk.3.post_ffw_norm_2.weight | Block 3 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 90 | blk.3.pre_ffw_norm_2.weight | Block 3 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.3: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.3: 2.5139 bits + + +### Block 4 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 91 | blk.4.attn_k.weight | Block 4 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 92 | blk.4.attn_k_norm.weight | Block 4 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 93 | blk.4.attn_norm.weight | Block 4 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 94 | blk.4.attn_output.weight | Block 4 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 95 | blk.4.attn_q.weight | Block 4 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 96 | blk.4.attn_q_norm.weight | Block 4 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 97 | blk.4.attn_v.weight | Block 4 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 98 | blk.4.ffn_down.weight | Block 4 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 99 | blk.4.ffn_down_exps.scale | Block 4 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 100 | blk.4.ffn_down_exps.weight | Block 4 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 101 | blk.4.ffn_gate.weight | Block 4 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 102 | blk.4.ffn_gate_inp.scale | Block 4 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 103 | blk.4.ffn_gate_inp.weight | Block 4 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 104 | blk.4.ffn_gate_up_exps.weight | Block 4 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 105 | blk.4.ffn_norm.weight | Block 4 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 106 | blk.4.ffn_up.weight | Block 4 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 107 | blk.4.layer_output_scale.weight | Block 4 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 108 | blk.4.post_attention_norm.weight | Block 4 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 109 | blk.4.post_ffw_norm.weight | Block 4 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 110 | blk.4.post_ffw_norm_1.weight | Block 4 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 111 | blk.4.post_ffw_norm_2.weight | Block 4 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 112 | blk.4.pre_ffw_norm_2.weight | Block 4 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.4: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.4: 2.5139 bits + + +### Block 5 Tensor Group : ~829M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 113 | blk.5.attn_k.weight | Block 5 Attention Key (W) | ( ~3M) 2883584 | 2816 x 1024 x 1 x 1 | IQ1_S | 1.5625 | +| 114 | blk.5.attn_k_norm.weight | Block 5 Attn_K_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 115 | blk.5.attn_norm.weight | Block 5 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 116 | blk.5.attn_output.weight | Block 5 Attention Output (W) | ( ~23M) 23068672 | 8192 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 117 | blk.5.attn_q.weight | Block 5 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | IQ1_S | 1.5625 | +| 118 | blk.5.attn_q_norm.weight | Block 5 Attn_Q_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 119 | blk.5.ffn_down.weight | Block 5 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 120 | blk.5.ffn_down_exps.scale | Block 5 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 121 | blk.5.ffn_down_exps.weight | Block 5 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 122 | blk.5.ffn_gate.weight | Block 5 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 123 | blk.5.ffn_gate_inp.scale | Block 5 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 124 | blk.5.ffn_gate_inp.weight | Block 5 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 125 | blk.5.ffn_gate_up_exps.weight | Block 5 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 126 | blk.5.ffn_norm.weight | Block 5 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 127 | blk.5.ffn_up.weight | Block 5 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 128 | blk.5.layer_output_scale.weight | Block 5 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 129 | blk.5.post_attention_norm.weight | Block 5 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 130 | blk.5.post_ffw_norm.weight | Block 5 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 131 | blk.5.post_ffw_norm_1.weight | Block 5 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 132 | blk.5.post_ffw_norm_2.weight | Block 5 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 133 | blk.5.pre_ffw_norm_2.weight | Block 5 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.5: (~829M) 828513409 +- Percentage of total elements: 3.28% +- Bits per Weight (BPW) for blk.5: 2.4974 bits + + +### Block 6 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 134 | blk.6.attn_k.weight | Block 6 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 135 | blk.6.attn_k_norm.weight | Block 6 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 136 | blk.6.attn_norm.weight | Block 6 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 137 | blk.6.attn_output.weight | Block 6 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 138 | blk.6.attn_q.weight | Block 6 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 139 | blk.6.attn_q_norm.weight | Block 6 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 140 | blk.6.attn_v.weight | Block 6 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 141 | blk.6.ffn_down.weight | Block 6 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 142 | blk.6.ffn_down_exps.scale | Block 6 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 143 | blk.6.ffn_down_exps.weight | Block 6 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 144 | blk.6.ffn_gate.weight | Block 6 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 145 | blk.6.ffn_gate_inp.scale | Block 6 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 146 | blk.6.ffn_gate_inp.weight | Block 6 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 147 | blk.6.ffn_gate_up_exps.weight | Block 6 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 148 | blk.6.ffn_norm.weight | Block 6 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 149 | blk.6.ffn_up.weight | Block 6 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 150 | blk.6.layer_output_scale.weight | Block 6 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 151 | blk.6.post_attention_norm.weight | Block 6 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 152 | blk.6.post_ffw_norm.weight | Block 6 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 153 | blk.6.post_ffw_norm_1.weight | Block 6 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 154 | blk.6.post_ffw_norm_2.weight | Block 6 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 155 | blk.6.pre_ffw_norm_2.weight | Block 6 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.6: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.6: 2.5139 bits + + +### Block 7 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 156 | blk.7.attn_k.weight | Block 7 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 157 | blk.7.attn_k_norm.weight | Block 7 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 158 | blk.7.attn_norm.weight | Block 7 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 159 | blk.7.attn_output.weight | Block 7 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 160 | blk.7.attn_q.weight | Block 7 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 161 | blk.7.attn_q_norm.weight | Block 7 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 162 | blk.7.attn_v.weight | Block 7 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 163 | blk.7.ffn_down.weight | Block 7 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 164 | blk.7.ffn_down_exps.scale | Block 7 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 165 | blk.7.ffn_down_exps.weight | Block 7 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 166 | blk.7.ffn_gate.weight | Block 7 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 167 | blk.7.ffn_gate_inp.scale | Block 7 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 168 | blk.7.ffn_gate_inp.weight | Block 7 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 169 | blk.7.ffn_gate_up_exps.weight | Block 7 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 170 | blk.7.ffn_norm.weight | Block 7 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 171 | blk.7.ffn_up.weight | Block 7 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 172 | blk.7.layer_output_scale.weight | Block 7 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 173 | blk.7.post_attention_norm.weight | Block 7 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 174 | blk.7.post_ffw_norm.weight | Block 7 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 175 | blk.7.post_ffw_norm_1.weight | Block 7 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 176 | blk.7.post_ffw_norm_2.weight | Block 7 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 177 | blk.7.pre_ffw_norm_2.weight | Block 7 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.7: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.7: 2.5139 bits + + +### Block 8 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 178 | blk.8.attn_k.weight | Block 8 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 179 | blk.8.attn_k_norm.weight | Block 8 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 180 | blk.8.attn_norm.weight | Block 8 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 181 | blk.8.attn_output.weight | Block 8 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 182 | blk.8.attn_q.weight | Block 8 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 183 | blk.8.attn_q_norm.weight | Block 8 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 184 | blk.8.attn_v.weight | Block 8 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 185 | blk.8.ffn_down.weight | Block 8 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 186 | blk.8.ffn_down_exps.scale | Block 8 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 187 | blk.8.ffn_down_exps.weight | Block 8 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 188 | blk.8.ffn_gate.weight | Block 8 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 189 | blk.8.ffn_gate_inp.scale | Block 8 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 190 | blk.8.ffn_gate_inp.weight | Block 8 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 191 | blk.8.ffn_gate_up_exps.weight | Block 8 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 192 | blk.8.ffn_norm.weight | Block 8 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 193 | blk.8.ffn_up.weight | Block 8 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 194 | blk.8.layer_output_scale.weight | Block 8 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 195 | blk.8.post_attention_norm.weight | Block 8 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 196 | blk.8.post_ffw_norm.weight | Block 8 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 197 | blk.8.post_ffw_norm_1.weight | Block 8 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 198 | blk.8.post_ffw_norm_2.weight | Block 8 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 199 | blk.8.pre_ffw_norm_2.weight | Block 8 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.8: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.8: 2.5139 bits + + +### Block 9 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :------------------------------- | :------------------------------------------------------------------------------------------ | :---------------- | :-------------------- | :----- | ------: | +| 200 | blk.9.attn_k.weight | Block 9 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 201 | blk.9.attn_k_norm.weight | Block 9 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 202 | blk.9.attn_norm.weight | Block 9 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 203 | blk.9.attn_output.weight | Block 9 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 204 | blk.9.attn_q.weight | Block 9 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 205 | blk.9.attn_q_norm.weight | Block 9 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 206 | blk.9.attn_v.weight | Block 9 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 207 | blk.9.ffn_down.weight | Block 9 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 208 | blk.9.ffn_down_exps.scale | Block 9 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 209 | blk.9.ffn_down_exps.weight | Block 9 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 210 | blk.9.ffn_gate.weight | Block 9 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 211 | blk.9.ffn_gate_inp.scale | Block 9 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 212 | blk.9.ffn_gate_inp.weight | Block 9 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 213 | blk.9.ffn_gate_up_exps.weight | Block 9 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 214 | blk.9.ffn_norm.weight | Block 9 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 215 | blk.9.ffn_up.weight | Block 9 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 216 | blk.9.layer_output_scale.weight | Block 9 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 217 | blk.9.post_attention_norm.weight | Block 9 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 218 | blk.9.post_ffw_norm.weight | Block 9 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 219 | blk.9.post_ffw_norm_1.weight | Block 9 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 220 | blk.9.post_ffw_norm_2.weight | Block 9 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 221 | blk.9.pre_ffw_norm_2.weight | Block 9 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.9: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.9: 2.5139 bits + + +### Block 10 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 222 | blk.10.attn_k.weight | Block 10 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 223 | blk.10.attn_k_norm.weight | Block 10 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 224 | blk.10.attn_norm.weight | Block 10 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 225 | blk.10.attn_output.weight | Block 10 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 226 | blk.10.attn_q.weight | Block 10 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 227 | blk.10.attn_q_norm.weight | Block 10 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 228 | blk.10.attn_v.weight | Block 10 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 229 | blk.10.ffn_down.weight | Block 10 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 230 | blk.10.ffn_down_exps.scale | Block 10 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 231 | blk.10.ffn_down_exps.weight | Block 10 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 232 | blk.10.ffn_gate.weight | Block 10 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 233 | blk.10.ffn_gate_inp.scale | Block 10 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 234 | blk.10.ffn_gate_inp.weight | Block 10 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 235 | blk.10.ffn_gate_up_exps.weight | Block 10 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 236 | blk.10.ffn_norm.weight | Block 10 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 237 | blk.10.ffn_up.weight | Block 10 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 238 | blk.10.layer_output_scale.weight | Block 10 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 239 | blk.10.post_attention_norm.weight | Block 10 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 240 | blk.10.post_ffw_norm.weight | Block 10 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 241 | blk.10.post_ffw_norm_1.weight | Block 10 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 242 | blk.10.post_ffw_norm_2.weight | Block 10 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 243 | blk.10.pre_ffw_norm_2.weight | Block 10 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.10: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.10: 2.5139 bits + + +### Block 11 Tensor Group : ~829M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 244 | blk.11.attn_k.weight | Block 11 Attention Key (W) | ( ~3M) 2883584 | 2816 x 1024 x 1 x 1 | IQ1_S | 1.5625 | +| 245 | blk.11.attn_k_norm.weight | Block 11 Attn_K_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 246 | blk.11.attn_norm.weight | Block 11 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 247 | blk.11.attn_output.weight | Block 11 Attention Output (W) | ( ~23M) 23068672 | 8192 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 248 | blk.11.attn_q.weight | Block 11 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | IQ1_S | 1.5625 | +| 249 | blk.11.attn_q_norm.weight | Block 11 Attn_Q_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 250 | blk.11.ffn_down.weight | Block 11 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 251 | blk.11.ffn_down_exps.scale | Block 11 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 252 | blk.11.ffn_down_exps.weight | Block 11 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 253 | blk.11.ffn_gate.weight | Block 11 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 254 | blk.11.ffn_gate_inp.scale | Block 11 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 255 | blk.11.ffn_gate_inp.weight | Block 11 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 256 | blk.11.ffn_gate_up_exps.weight | Block 11 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 257 | blk.11.ffn_norm.weight | Block 11 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 258 | blk.11.ffn_up.weight | Block 11 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 259 | blk.11.layer_output_scale.weight | Block 11 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 260 | blk.11.post_attention_norm.weight | Block 11 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 261 | blk.11.post_ffw_norm.weight | Block 11 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 262 | blk.11.post_ffw_norm_1.weight | Block 11 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 263 | blk.11.post_ffw_norm_2.weight | Block 11 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 264 | blk.11.pre_ffw_norm_2.weight | Block 11 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.11: (~829M) 828513409 +- Percentage of total elements: 3.28% +- Bits per Weight (BPW) for blk.11: 2.4974 bits + + +### Block 12 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 265 | blk.12.attn_k.weight | Block 12 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 266 | blk.12.attn_k_norm.weight | Block 12 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 267 | blk.12.attn_norm.weight | Block 12 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 268 | blk.12.attn_output.weight | Block 12 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 269 | blk.12.attn_q.weight | Block 12 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 270 | blk.12.attn_q_norm.weight | Block 12 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 271 | blk.12.attn_v.weight | Block 12 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 272 | blk.12.ffn_down.weight | Block 12 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 273 | blk.12.ffn_down_exps.scale | Block 12 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 274 | blk.12.ffn_down_exps.weight | Block 12 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 275 | blk.12.ffn_gate.weight | Block 12 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 276 | blk.12.ffn_gate_inp.scale | Block 12 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 277 | blk.12.ffn_gate_inp.weight | Block 12 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 278 | blk.12.ffn_gate_up_exps.weight | Block 12 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 279 | blk.12.ffn_norm.weight | Block 12 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 280 | blk.12.ffn_up.weight | Block 12 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 281 | blk.12.layer_output_scale.weight | Block 12 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 282 | blk.12.post_attention_norm.weight | Block 12 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 283 | blk.12.post_ffw_norm.weight | Block 12 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 284 | blk.12.post_ffw_norm_1.weight | Block 12 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 285 | blk.12.post_ffw_norm_2.weight | Block 12 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 286 | blk.12.pre_ffw_norm_2.weight | Block 12 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.12: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.12: 2.5139 bits + + +### Block 13 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 287 | blk.13.attn_k.weight | Block 13 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 288 | blk.13.attn_k_norm.weight | Block 13 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 289 | blk.13.attn_norm.weight | Block 13 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 290 | blk.13.attn_output.weight | Block 13 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 291 | blk.13.attn_q.weight | Block 13 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 292 | blk.13.attn_q_norm.weight | Block 13 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 293 | blk.13.attn_v.weight | Block 13 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 294 | blk.13.ffn_down.weight | Block 13 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 295 | blk.13.ffn_down_exps.scale | Block 13 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 296 | blk.13.ffn_down_exps.weight | Block 13 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 297 | blk.13.ffn_gate.weight | Block 13 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 298 | blk.13.ffn_gate_inp.scale | Block 13 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 299 | blk.13.ffn_gate_inp.weight | Block 13 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 300 | blk.13.ffn_gate_up_exps.weight | Block 13 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 301 | blk.13.ffn_norm.weight | Block 13 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 302 | blk.13.ffn_up.weight | Block 13 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 303 | blk.13.layer_output_scale.weight | Block 13 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 304 | blk.13.post_attention_norm.weight | Block 13 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 305 | blk.13.post_ffw_norm.weight | Block 13 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 306 | blk.13.post_ffw_norm_1.weight | Block 13 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 307 | blk.13.post_ffw_norm_2.weight | Block 13 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 308 | blk.13.pre_ffw_norm_2.weight | Block 13 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.13: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.13: 2.5139 bits + + +### Block 14 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 309 | blk.14.attn_k.weight | Block 14 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 310 | blk.14.attn_k_norm.weight | Block 14 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 311 | blk.14.attn_norm.weight | Block 14 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 312 | blk.14.attn_output.weight | Block 14 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 313 | blk.14.attn_q.weight | Block 14 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 314 | blk.14.attn_q_norm.weight | Block 14 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 315 | blk.14.attn_v.weight | Block 14 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 316 | blk.14.ffn_down.weight | Block 14 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 317 | blk.14.ffn_down_exps.scale | Block 14 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 318 | blk.14.ffn_down_exps.weight | Block 14 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 319 | blk.14.ffn_gate.weight | Block 14 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 320 | blk.14.ffn_gate_inp.scale | Block 14 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 321 | blk.14.ffn_gate_inp.weight | Block 14 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 322 | blk.14.ffn_gate_up_exps.weight | Block 14 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 323 | blk.14.ffn_norm.weight | Block 14 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 324 | blk.14.ffn_up.weight | Block 14 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 325 | blk.14.layer_output_scale.weight | Block 14 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 326 | blk.14.post_attention_norm.weight | Block 14 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 327 | blk.14.post_ffw_norm.weight | Block 14 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 328 | blk.14.post_ffw_norm_1.weight | Block 14 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 329 | blk.14.post_ffw_norm_2.weight | Block 14 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 330 | blk.14.pre_ffw_norm_2.weight | Block 14 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.14: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.14: 2.5139 bits + + +### Block 15 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 331 | blk.15.attn_k.weight | Block 15 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 332 | blk.15.attn_k_norm.weight | Block 15 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 333 | blk.15.attn_norm.weight | Block 15 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 334 | blk.15.attn_output.weight | Block 15 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 335 | blk.15.attn_q.weight | Block 15 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 336 | blk.15.attn_q_norm.weight | Block 15 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 337 | blk.15.attn_v.weight | Block 15 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 338 | blk.15.ffn_down.weight | Block 15 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 339 | blk.15.ffn_down_exps.scale | Block 15 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 340 | blk.15.ffn_down_exps.weight | Block 15 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 341 | blk.15.ffn_gate.weight | Block 15 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 342 | blk.15.ffn_gate_inp.scale | Block 15 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 343 | blk.15.ffn_gate_inp.weight | Block 15 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 344 | blk.15.ffn_gate_up_exps.weight | Block 15 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 345 | blk.15.ffn_norm.weight | Block 15 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 346 | blk.15.ffn_up.weight | Block 15 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 347 | blk.15.layer_output_scale.weight | Block 15 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 348 | blk.15.post_attention_norm.weight | Block 15 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 349 | blk.15.post_ffw_norm.weight | Block 15 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 350 | blk.15.post_ffw_norm_1.weight | Block 15 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 351 | blk.15.post_ffw_norm_2.weight | Block 15 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 352 | blk.15.pre_ffw_norm_2.weight | Block 15 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.15: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.15: 2.5139 bits + + +### Block 16 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 353 | blk.16.attn_k.weight | Block 16 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 354 | blk.16.attn_k_norm.weight | Block 16 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 355 | blk.16.attn_norm.weight | Block 16 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 356 | blk.16.attn_output.weight | Block 16 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 357 | blk.16.attn_q.weight | Block 16 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 358 | blk.16.attn_q_norm.weight | Block 16 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 359 | blk.16.attn_v.weight | Block 16 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 360 | blk.16.ffn_down.weight | Block 16 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 361 | blk.16.ffn_down_exps.scale | Block 16 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 362 | blk.16.ffn_down_exps.weight | Block 16 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 363 | blk.16.ffn_gate.weight | Block 16 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 364 | blk.16.ffn_gate_inp.scale | Block 16 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 365 | blk.16.ffn_gate_inp.weight | Block 16 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 366 | blk.16.ffn_gate_up_exps.weight | Block 16 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 367 | blk.16.ffn_norm.weight | Block 16 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 368 | blk.16.ffn_up.weight | Block 16 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 369 | blk.16.layer_output_scale.weight | Block 16 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 370 | blk.16.post_attention_norm.weight | Block 16 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 371 | blk.16.post_ffw_norm.weight | Block 16 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 372 | blk.16.post_ffw_norm_1.weight | Block 16 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 373 | blk.16.post_ffw_norm_2.weight | Block 16 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 374 | blk.16.pre_ffw_norm_2.weight | Block 16 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.16: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.16: 2.5139 bits + + +### Block 17 Tensor Group : ~829M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 375 | blk.17.attn_k.weight | Block 17 Attention Key (W) | ( ~3M) 2883584 | 2816 x 1024 x 1 x 1 | IQ1_S | 1.5625 | +| 376 | blk.17.attn_k_norm.weight | Block 17 Attn_K_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 377 | blk.17.attn_norm.weight | Block 17 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 378 | blk.17.attn_output.weight | Block 17 Attention Output (W) | ( ~23M) 23068672 | 8192 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 379 | blk.17.attn_q.weight | Block 17 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | IQ1_S | 1.5625 | +| 380 | blk.17.attn_q_norm.weight | Block 17 Attn_Q_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 381 | blk.17.ffn_down.weight | Block 17 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 382 | blk.17.ffn_down_exps.scale | Block 17 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 383 | blk.17.ffn_down_exps.weight | Block 17 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 384 | blk.17.ffn_gate.weight | Block 17 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 385 | blk.17.ffn_gate_inp.scale | Block 17 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 386 | blk.17.ffn_gate_inp.weight | Block 17 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 387 | blk.17.ffn_gate_up_exps.weight | Block 17 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 388 | blk.17.ffn_norm.weight | Block 17 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 389 | blk.17.ffn_up.weight | Block 17 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 390 | blk.17.layer_output_scale.weight | Block 17 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 391 | blk.17.post_attention_norm.weight | Block 17 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 392 | blk.17.post_ffw_norm.weight | Block 17 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 393 | blk.17.post_ffw_norm_1.weight | Block 17 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 394 | blk.17.post_ffw_norm_2.weight | Block 17 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 395 | blk.17.pre_ffw_norm_2.weight | Block 17 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.17: (~829M) 828513409 +- Percentage of total elements: 3.28% +- Bits per Weight (BPW) for blk.17: 2.4974 bits + + +### Block 18 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 396 | blk.18.attn_k.weight | Block 18 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 397 | blk.18.attn_k_norm.weight | Block 18 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 398 | blk.18.attn_norm.weight | Block 18 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 399 | blk.18.attn_output.weight | Block 18 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 400 | blk.18.attn_q.weight | Block 18 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 401 | blk.18.attn_q_norm.weight | Block 18 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 402 | blk.18.attn_v.weight | Block 18 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 403 | blk.18.ffn_down.weight | Block 18 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 404 | blk.18.ffn_down_exps.scale | Block 18 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 405 | blk.18.ffn_down_exps.weight | Block 18 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 406 | blk.18.ffn_gate.weight | Block 18 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 407 | blk.18.ffn_gate_inp.scale | Block 18 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 408 | blk.18.ffn_gate_inp.weight | Block 18 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 409 | blk.18.ffn_gate_up_exps.weight | Block 18 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 410 | blk.18.ffn_norm.weight | Block 18 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 411 | blk.18.ffn_up.weight | Block 18 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 412 | blk.18.layer_output_scale.weight | Block 18 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 413 | blk.18.post_attention_norm.weight | Block 18 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 414 | blk.18.post_ffw_norm.weight | Block 18 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 415 | blk.18.post_ffw_norm_1.weight | Block 18 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 416 | blk.18.post_ffw_norm_2.weight | Block 18 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 417 | blk.18.pre_ffw_norm_2.weight | Block 18 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.18: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.18: 2.5139 bits + + +### Block 19 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 418 | blk.19.attn_k.weight | Block 19 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 419 | blk.19.attn_k_norm.weight | Block 19 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 420 | blk.19.attn_norm.weight | Block 19 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 421 | blk.19.attn_output.weight | Block 19 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 422 | blk.19.attn_q.weight | Block 19 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 423 | blk.19.attn_q_norm.weight | Block 19 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 424 | blk.19.attn_v.weight | Block 19 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 425 | blk.19.ffn_down.weight | Block 19 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 426 | blk.19.ffn_down_exps.scale | Block 19 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 427 | blk.19.ffn_down_exps.weight | Block 19 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 428 | blk.19.ffn_gate.weight | Block 19 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 429 | blk.19.ffn_gate_inp.scale | Block 19 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 430 | blk.19.ffn_gate_inp.weight | Block 19 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 431 | blk.19.ffn_gate_up_exps.weight | Block 19 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 432 | blk.19.ffn_norm.weight | Block 19 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 433 | blk.19.ffn_up.weight | Block 19 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 434 | blk.19.layer_output_scale.weight | Block 19 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 435 | blk.19.post_attention_norm.weight | Block 19 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 436 | blk.19.post_ffw_norm.weight | Block 19 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 437 | blk.19.post_ffw_norm_1.weight | Block 19 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 438 | blk.19.post_ffw_norm_2.weight | Block 19 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 439 | blk.19.pre_ffw_norm_2.weight | Block 19 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.19: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.19: 2.5139 bits + + +### Block 20 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 440 | blk.20.attn_k.weight | Block 20 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 441 | blk.20.attn_k_norm.weight | Block 20 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 442 | blk.20.attn_norm.weight | Block 20 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 443 | blk.20.attn_output.weight | Block 20 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 444 | blk.20.attn_q.weight | Block 20 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 445 | blk.20.attn_q_norm.weight | Block 20 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 446 | blk.20.attn_v.weight | Block 20 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 447 | blk.20.ffn_down.weight | Block 20 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 448 | blk.20.ffn_down_exps.scale | Block 20 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 449 | blk.20.ffn_down_exps.weight | Block 20 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 450 | blk.20.ffn_gate.weight | Block 20 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 451 | blk.20.ffn_gate_inp.scale | Block 20 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 452 | blk.20.ffn_gate_inp.weight | Block 20 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 453 | blk.20.ffn_gate_up_exps.weight | Block 20 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 454 | blk.20.ffn_norm.weight | Block 20 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 455 | blk.20.ffn_up.weight | Block 20 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 456 | blk.20.layer_output_scale.weight | Block 20 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 457 | blk.20.post_attention_norm.weight | Block 20 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 458 | blk.20.post_ffw_norm.weight | Block 20 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 459 | blk.20.post_ffw_norm_1.weight | Block 20 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 460 | blk.20.post_ffw_norm_2.weight | Block 20 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 461 | blk.20.pre_ffw_norm_2.weight | Block 20 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.20: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.20: 2.5139 bits + + +### Block 21 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 462 | blk.21.attn_k.weight | Block 21 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 463 | blk.21.attn_k_norm.weight | Block 21 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 464 | blk.21.attn_norm.weight | Block 21 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 465 | blk.21.attn_output.weight | Block 21 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 466 | blk.21.attn_q.weight | Block 21 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 467 | blk.21.attn_q_norm.weight | Block 21 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 468 | blk.21.attn_v.weight | Block 21 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 469 | blk.21.ffn_down.weight | Block 21 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 470 | blk.21.ffn_down_exps.scale | Block 21 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 471 | blk.21.ffn_down_exps.weight | Block 21 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 472 | blk.21.ffn_gate.weight | Block 21 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 473 | blk.21.ffn_gate_inp.scale | Block 21 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 474 | blk.21.ffn_gate_inp.weight | Block 21 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 475 | blk.21.ffn_gate_up_exps.weight | Block 21 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 476 | blk.21.ffn_norm.weight | Block 21 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 477 | blk.21.ffn_up.weight | Block 21 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 478 | blk.21.layer_output_scale.weight | Block 21 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 479 | blk.21.post_attention_norm.weight | Block 21 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 480 | blk.21.post_ffw_norm.weight | Block 21 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 481 | blk.21.post_ffw_norm_1.weight | Block 21 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 482 | blk.21.post_ffw_norm_2.weight | Block 21 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 483 | blk.21.pre_ffw_norm_2.weight | Block 21 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.21: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.21: 2.5139 bits + + +### Block 22 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 484 | blk.22.attn_k.weight | Block 22 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 485 | blk.22.attn_k_norm.weight | Block 22 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 486 | blk.22.attn_norm.weight | Block 22 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 487 | blk.22.attn_output.weight | Block 22 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 488 | blk.22.attn_q.weight | Block 22 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 489 | blk.22.attn_q_norm.weight | Block 22 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 490 | blk.22.attn_v.weight | Block 22 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 491 | blk.22.ffn_down.weight | Block 22 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 492 | blk.22.ffn_down_exps.scale | Block 22 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 493 | blk.22.ffn_down_exps.weight | Block 22 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 494 | blk.22.ffn_gate.weight | Block 22 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 495 | blk.22.ffn_gate_inp.scale | Block 22 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 496 | blk.22.ffn_gate_inp.weight | Block 22 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 497 | blk.22.ffn_gate_up_exps.weight | Block 22 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 498 | blk.22.ffn_norm.weight | Block 22 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 499 | blk.22.ffn_up.weight | Block 22 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 500 | blk.22.layer_output_scale.weight | Block 22 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 501 | blk.22.post_attention_norm.weight | Block 22 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 502 | blk.22.post_ffw_norm.weight | Block 22 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 503 | blk.22.post_ffw_norm_1.weight | Block 22 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 504 | blk.22.post_ffw_norm_2.weight | Block 22 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 505 | blk.22.pre_ffw_norm_2.weight | Block 22 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.22: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.22: 2.5139 bits + + +### Block 23 Tensor Group : ~829M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 506 | blk.23.attn_k.weight | Block 23 Attention Key (W) | ( ~3M) 2883584 | 2816 x 1024 x 1 x 1 | IQ1_S | 1.5625 | +| 507 | blk.23.attn_k_norm.weight | Block 23 Attn_K_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 508 | blk.23.attn_norm.weight | Block 23 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 509 | blk.23.attn_output.weight | Block 23 Attention Output (W) | ( ~23M) 23068672 | 8192 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 510 | blk.23.attn_q.weight | Block 23 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | IQ1_S | 1.5625 | +| 511 | blk.23.attn_q_norm.weight | Block 23 Attn_Q_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 512 | blk.23.ffn_down.weight | Block 23 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 513 | blk.23.ffn_down_exps.scale | Block 23 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 514 | blk.23.ffn_down_exps.weight | Block 23 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 515 | blk.23.ffn_gate.weight | Block 23 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 516 | blk.23.ffn_gate_inp.scale | Block 23 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 517 | blk.23.ffn_gate_inp.weight | Block 23 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 518 | blk.23.ffn_gate_up_exps.weight | Block 23 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 519 | blk.23.ffn_norm.weight | Block 23 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 520 | blk.23.ffn_up.weight | Block 23 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 521 | blk.23.layer_output_scale.weight | Block 23 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 522 | blk.23.post_attention_norm.weight | Block 23 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 523 | blk.23.post_ffw_norm.weight | Block 23 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 524 | blk.23.post_ffw_norm_1.weight | Block 23 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 525 | blk.23.post_ffw_norm_2.weight | Block 23 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 526 | blk.23.pre_ffw_norm_2.weight | Block 23 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.23: (~829M) 828513409 +- Percentage of total elements: 3.28% +- Bits per Weight (BPW) for blk.23: 2.4974 bits + + +### Block 24 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 527 | blk.24.attn_k.weight | Block 24 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 528 | blk.24.attn_k_norm.weight | Block 24 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 529 | blk.24.attn_norm.weight | Block 24 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 530 | blk.24.attn_output.weight | Block 24 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 531 | blk.24.attn_q.weight | Block 24 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 532 | blk.24.attn_q_norm.weight | Block 24 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 533 | blk.24.attn_v.weight | Block 24 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 534 | blk.24.ffn_down.weight | Block 24 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 535 | blk.24.ffn_down_exps.scale | Block 24 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 536 | blk.24.ffn_down_exps.weight | Block 24 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 537 | blk.24.ffn_gate.weight | Block 24 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 538 | blk.24.ffn_gate_inp.scale | Block 24 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 539 | blk.24.ffn_gate_inp.weight | Block 24 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 540 | blk.24.ffn_gate_up_exps.weight | Block 24 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 541 | blk.24.ffn_norm.weight | Block 24 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 542 | blk.24.ffn_up.weight | Block 24 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 543 | blk.24.layer_output_scale.weight | Block 24 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 544 | blk.24.post_attention_norm.weight | Block 24 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 545 | blk.24.post_ffw_norm.weight | Block 24 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 546 | blk.24.post_ffw_norm_1.weight | Block 24 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 547 | blk.24.post_ffw_norm_2.weight | Block 24 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 548 | blk.24.pre_ffw_norm_2.weight | Block 24 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.24: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.24: 2.5139 bits + + +### Block 25 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 549 | blk.25.attn_k.weight | Block 25 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 550 | blk.25.attn_k_norm.weight | Block 25 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 551 | blk.25.attn_norm.weight | Block 25 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 552 | blk.25.attn_output.weight | Block 25 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 553 | blk.25.attn_q.weight | Block 25 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 554 | blk.25.attn_q_norm.weight | Block 25 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 555 | blk.25.attn_v.weight | Block 25 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 556 | blk.25.ffn_down.weight | Block 25 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 557 | blk.25.ffn_down_exps.scale | Block 25 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 558 | blk.25.ffn_down_exps.weight | Block 25 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 559 | blk.25.ffn_gate.weight | Block 25 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 560 | blk.25.ffn_gate_inp.scale | Block 25 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 561 | blk.25.ffn_gate_inp.weight | Block 25 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 562 | blk.25.ffn_gate_up_exps.weight | Block 25 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 563 | blk.25.ffn_norm.weight | Block 25 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 564 | blk.25.ffn_up.weight | Block 25 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 565 | blk.25.layer_output_scale.weight | Block 25 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 566 | blk.25.post_attention_norm.weight | Block 25 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 567 | blk.25.post_ffw_norm.weight | Block 25 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 568 | blk.25.post_ffw_norm_1.weight | Block 25 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 569 | blk.25.post_ffw_norm_2.weight | Block 25 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 570 | blk.25.pre_ffw_norm_2.weight | Block 25 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.25: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.25: 2.5139 bits + + +### Block 26 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 571 | blk.26.attn_k.weight | Block 26 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 572 | blk.26.attn_k_norm.weight | Block 26 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 573 | blk.26.attn_norm.weight | Block 26 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 574 | blk.26.attn_output.weight | Block 26 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 575 | blk.26.attn_q.weight | Block 26 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 576 | blk.26.attn_q_norm.weight | Block 26 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 577 | blk.26.attn_v.weight | Block 26 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 578 | blk.26.ffn_down.weight | Block 26 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 579 | blk.26.ffn_down_exps.scale | Block 26 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 580 | blk.26.ffn_down_exps.weight | Block 26 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 581 | blk.26.ffn_gate.weight | Block 26 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 582 | blk.26.ffn_gate_inp.scale | Block 26 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 583 | blk.26.ffn_gate_inp.weight | Block 26 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 584 | blk.26.ffn_gate_up_exps.weight | Block 26 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 585 | blk.26.ffn_norm.weight | Block 26 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 586 | blk.26.ffn_up.weight | Block 26 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 587 | blk.26.layer_output_scale.weight | Block 26 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 588 | blk.26.post_attention_norm.weight | Block 26 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 589 | blk.26.post_ffw_norm.weight | Block 26 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 590 | blk.26.post_ffw_norm_1.weight | Block 26 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 591 | blk.26.post_ffw_norm_2.weight | Block 26 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 592 | blk.26.pre_ffw_norm_2.weight | Block 26 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.26: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.26: 2.5139 bits + + +### Block 27 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 593 | blk.27.attn_k.weight | Block 27 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 594 | blk.27.attn_k_norm.weight | Block 27 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 595 | blk.27.attn_norm.weight | Block 27 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 596 | blk.27.attn_output.weight | Block 27 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 597 | blk.27.attn_q.weight | Block 27 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 598 | blk.27.attn_q_norm.weight | Block 27 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 599 | blk.27.attn_v.weight | Block 27 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 600 | blk.27.ffn_down.weight | Block 27 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 601 | blk.27.ffn_down_exps.scale | Block 27 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 602 | blk.27.ffn_down_exps.weight | Block 27 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 603 | blk.27.ffn_gate.weight | Block 27 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 604 | blk.27.ffn_gate_inp.scale | Block 27 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 605 | blk.27.ffn_gate_inp.weight | Block 27 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 606 | blk.27.ffn_gate_up_exps.weight | Block 27 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 607 | blk.27.ffn_norm.weight | Block 27 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 608 | blk.27.ffn_up.weight | Block 27 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 609 | blk.27.layer_output_scale.weight | Block 27 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 610 | blk.27.post_attention_norm.weight | Block 27 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 611 | blk.27.post_ffw_norm.weight | Block 27 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 612 | blk.27.post_ffw_norm_1.weight | Block 27 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 613 | blk.27.post_ffw_norm_2.weight | Block 27 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 614 | blk.27.pre_ffw_norm_2.weight | Block 27 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.27: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.27: 2.5139 bits + + +### Block 28 Tensor Group : ~814M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 615 | blk.28.attn_k.weight | Block 28 Attention Key (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 616 | blk.28.attn_k_norm.weight | Block 28 Attn_K_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 617 | blk.28.attn_norm.weight | Block 28 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 618 | blk.28.attn_output.weight | Block 28 Attention Output (W) | ( ~12M) 11534336 | 4096 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 619 | blk.28.attn_q.weight | Block 28 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | IQ1_S | 1.5625 | +| 620 | blk.28.attn_q_norm.weight | Block 28 Attn_Q_Norm (W) | ( 256) 256 | 256 x 1 x 1 x 1 | F32 | 32.0000 | +| 621 | blk.28.attn_v.weight | Block 28 Attention Value (W) | ( ~6M) 5767168 | 2816 x 2048 x 1 x 1 | IQ1_S | 1.5625 | +| 622 | blk.28.ffn_down.weight | Block 28 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 623 | blk.28.ffn_down_exps.scale | Block 28 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 624 | blk.28.ffn_down_exps.weight | Block 28 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 625 | blk.28.ffn_gate.weight | Block 28 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 626 | blk.28.ffn_gate_inp.scale | Block 28 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 627 | blk.28.ffn_gate_inp.weight | Block 28 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 628 | blk.28.ffn_gate_up_exps.weight | Block 28 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 629 | blk.28.ffn_norm.weight | Block 28 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 630 | blk.28.ffn_up.weight | Block 28 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 631 | blk.28.layer_output_scale.weight | Block 28 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 632 | blk.28.post_attention_norm.weight | Block 28 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 633 | blk.28.post_ffw_norm.weight | Block 28 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 634 | blk.28.post_ffw_norm_1.weight | Block 28 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 635 | blk.28.post_ffw_norm_2.weight | Block 28 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 636 | blk.28.pre_ffw_norm_2.weight | Block 28 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.28: (~814M) 814094977 +- Percentage of total elements: 3.23% +- Bits per Weight (BPW) for blk.28: 2.5139 bits + + +### Block 29 Tensor Group : ~829M Elements + +| T_ID | Tensor Layer Name | Human Friendly Tensor Layer Name | Elements | Shape | Type | BPW | +| ---: | :-------------------------------- | :------------------------------------------------------------------------------------------- | :---------------- | :-------------------- | :----- | ------: | +| 637 | blk.29.attn_k.weight | Block 29 Attention Key (W) | ( ~3M) 2883584 | 2816 x 1024 x 1 x 1 | IQ1_S | 1.5625 | +| 638 | blk.29.attn_k_norm.weight | Block 29 Attn_K_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 639 | blk.29.attn_norm.weight | Block 29 Attention Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 640 | blk.29.attn_output.weight | Block 29 Attention Output (W) | ( ~23M) 23068672 | 8192 x 2816 x 1 x 1 | IQ1_S | 1.5625 | +| 641 | blk.29.attn_q.weight | Block 29 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | IQ1_S | 1.5625 | +| 642 | blk.29.attn_q_norm.weight | Block 29 Attn_Q_Norm (W) | ( 512) 512 | 512 x 1 x 1 x 1 | F32 | 32.0000 | +| 643 | blk.29.ffn_down.weight | Block 29 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | IQ4_NL | 4.5000 | +| 644 | blk.29.ffn_down_exps.scale | Block 29 Ffn_Down_Exps Scale | ( 128) 128 | 128 x 1 x 1 x 1 | F32 | 32.0000 | +| 645 | blk.29.ffn_down_exps.weight | Block 29 Ffn_Down_Exps (W) | (~254M) 253755392 | 704 x 2816 x 128 x 1 | IQ4_NL | 4.5000 | +| 646 | blk.29.ffn_gate.weight | Block 29 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 647 | blk.29.ffn_gate_inp.scale | Block 29 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models Scale | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 648 | blk.29.ffn_gate_inp.weight | Block 29 Expert-Routing Layer For The Feed-Forward Network In Mixture Of Expert Models (W) | (~360K) 360448 | 2816 x 128 x 1 x 1 | F32 | 32.0000 | +| 649 | blk.29.ffn_gate_up_exps.weight | Block 29 Ffn_Gate_Up_Exps (W) | (~508M) 507510784 | 2816 x 1408 x 128 x 1 | IQ1_S | 1.5625 | +| 650 | blk.29.ffn_norm.weight | Block 29 Feed-Forward Network Normalization (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 651 | blk.29.ffn_up.weight | Block 29 Feed-Forward Network "Up" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | IQ1_S | 1.5625 | +| 652 | blk.29.layer_output_scale.weight | Block 29 Layer_Output_Scale (W) | ( 1) 1 | 1 x 1 x 1 x 1 | F32 | 32.0000 | +| 653 | blk.29.post_attention_norm.weight | Block 29 Post_Attention_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 654 | blk.29.post_ffw_norm.weight | Block 29 Post_Ffw_Norm (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 655 | blk.29.post_ffw_norm_1.weight | Block 29 Post_Ffw_Norm_1 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 656 | blk.29.post_ffw_norm_2.weight | Block 29 Post_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | +| 657 | blk.29.pre_ffw_norm_2.weight | Block 29 Pre_Ffw_Norm_2 (W) | ( ~3K) 2816 | 2816 x 1 x 1 x 1 | F32 | 32.0000 | + +- Total elements in blk.29: (~829M) 828513409 +- Percentage of total elements: 3.28% +- Bits per Weight (BPW) for blk.29: 2.4974 bits + + +Total BPW for gemma-4-26B-A4B-it-Q2_K.gguf: 2.5145 bits