diff --git "a/scores/gemma-4-26B-A4B-it-Q8_0.md" "b/scores/gemma-4-26B-A4B-it-Q8_0.md" new file mode 100644--- /dev/null +++ "b/scores/gemma-4-26B-A4B-it-Q8_0.md" @@ -0,0 +1,1742 @@ +# gemma-4-26B-A4B-it-Q8_0.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 | 7 | +| 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-Q8\_0.gguf - GGUF Internal File Dump](#userseddevelopmentaihfgemma-4-26b-a4b-it-wipgemma-4-26b-a4b-it-q8_0gguf---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 | 0x2ec00000 | +| 3 | blk.0.attn_k.weight | 0x2fb19c00 | 0x5d8000 | +| 4 | blk.0.attn_k_norm.weight | 0x300f1c00 | 0x400 | +| 5 | blk.0.attn_norm.weight | 0x300f2000 | 0x2c00 | +| 6 | blk.0.attn_output.weight | 0x300f4c00 | 0xbb0000 | +| 7 | blk.0.attn_q.weight | 0x30ca4c00 | 0xbb0000 | +| 8 | blk.0.attn_q_norm.weight | 0x31854c00 | 0x400 | +| 9 | blk.0.attn_v.weight | 0x31855000 | 0x5d8000 | +| 10 | blk.0.ffn_down.weight | 0x31e2d000 | 0x606c00 | +| 11 | blk.0.ffn_down_exps.scale | 0x32433c00 | 0x200 | +| 12 | blk.0.ffn_down_exps.weight | 0x32433e00 | 0x10120000 | +| 13 | blk.0.ffn_gate.weight | 0x42553e00 | 0x606c00 | +| 14 | blk.0.ffn_gate_inp.scale | 0x42b5aa00 | 0x2c00 | +| 15 | blk.0.ffn_gate_inp.weight | 0x42b5d600 | 0x160000 | +| 16 | blk.0.ffn_gate_up_exps.weight | 0x42cbd600 | 0x18d08000 | +| 17 | blk.0.ffn_norm.weight | 0x5b9c5600 | 0x2c00 | +| 18 | blk.0.ffn_up.weight | 0x5b9c8200 | 0x606c00 | +| 19 | blk.0.layer_output_scale.weight | 0x5bfcee00 | 0x4 | +| 20 | blk.0.post_attention_norm.weight | 0x5bfcee20 | 0x2c00 | +| 21 | blk.0.post_ffw_norm.weight | 0x5bfd1a20 | 0x2c00 | +| 22 | blk.0.post_ffw_norm_1.weight | 0x5bfd4620 | 0x2c00 | +| 23 | blk.0.post_ffw_norm_2.weight | 0x5bfd7220 | 0x2c00 | +| 24 | blk.0.pre_ffw_norm_2.weight | 0x5bfd9e20 | 0x2c00 | +| 25 | blk.1.attn_k.weight | 0x5bfdca20 | 0x5d8000 | +| 26 | blk.1.attn_k_norm.weight | 0x5c5b4a20 | 0x400 | +| 27 | blk.1.attn_norm.weight | 0x5c5b4e20 | 0x2c00 | +| 28 | blk.1.attn_output.weight | 0x5c5b7a20 | 0xbb0000 | +| 29 | blk.1.attn_q.weight | 0x5d167a20 | 0xbb0000 | +| 30 | blk.1.attn_q_norm.weight | 0x5dd17a20 | 0x400 | +| 31 | blk.1.attn_v.weight | 0x5dd17e20 | 0x5d8000 | +| 32 | blk.1.ffn_down.weight | 0x5e2efe20 | 0x606c00 | +| 33 | blk.1.ffn_down_exps.scale | 0x5e8f6a20 | 0x200 | +| 34 | blk.1.ffn_down_exps.weight | 0x5e8f6c20 | 0x10120000 | +| 35 | blk.1.ffn_gate.weight | 0x6ea16c20 | 0x606c00 | +| 36 | blk.1.ffn_gate_inp.scale | 0x6f01d820 | 0x2c00 | +| 37 | blk.1.ffn_gate_inp.weight | 0x6f020420 | 0x160000 | +| 38 | blk.1.ffn_gate_up_exps.weight | 0x6f180420 | 0x20240000 | +| 39 | blk.1.ffn_norm.weight | 0x8f3c0420 | 0x2c00 | +| 40 | blk.1.ffn_up.weight | 0x8f3c3020 | 0x606c00 | +| 41 | blk.1.layer_output_scale.weight | 0x8f9c9c20 | 0x4 | +| 42 | blk.1.post_attention_norm.weight | 0x8f9c9c40 | 0x2c00 | +| 43 | blk.1.post_ffw_norm.weight | 0x8f9cc840 | 0x2c00 | +| 44 | blk.1.post_ffw_norm_1.weight | 0x8f9cf440 | 0x2c00 | +| 45 | blk.1.post_ffw_norm_2.weight | 0x8f9d2040 | 0x2c00 | +| 46 | blk.1.pre_ffw_norm_2.weight | 0x8f9d4c40 | 0x2c00 | +| 47 | blk.2.attn_k.weight | 0x8f9d7840 | 0x5d8000 | +| 48 | blk.2.attn_k_norm.weight | 0x8ffaf840 | 0x400 | +| 49 | blk.2.attn_norm.weight | 0x8ffafc40 | 0x2c00 | +| 50 | blk.2.attn_output.weight | 0x8ffb2840 | 0xbb0000 | +| 51 | blk.2.attn_q.weight | 0x90b62840 | 0xbb0000 | +| 52 | blk.2.attn_q_norm.weight | 0x91712840 | 0x400 | +| 53 | blk.2.attn_v.weight | 0x91712c40 | 0x5d8000 | +| 54 | blk.2.ffn_down.weight | 0x91ceac40 | 0x606c00 | +| 55 | blk.2.ffn_down_exps.scale | 0x922f1840 | 0x200 | +| 56 | blk.2.ffn_down_exps.weight | 0x922f1a40 | 0x10120000 | +| 57 | blk.2.ffn_gate.weight | 0xa2411a40 | 0x606c00 | +| 58 | blk.2.ffn_gate_inp.scale | 0xa2a18640 | 0x2c00 | +| 59 | blk.2.ffn_gate_inp.weight | 0xa2a1b240 | 0x160000 | +| 60 | blk.2.ffn_gate_up_exps.weight | 0xa2b7b240 | 0x20240000 | +| 61 | blk.2.ffn_norm.weight | 0xc2dbb240 | 0x2c00 | +| 62 | blk.2.ffn_up.weight | 0xc2dbde40 | 0x606c00 | +| 63 | blk.2.layer_output_scale.weight | 0xc33c4a40 | 0x4 | +| 64 | blk.2.post_attention_norm.weight | 0xc33c4a60 | 0x2c00 | +| 65 | blk.2.post_ffw_norm.weight | 0xc33c7660 | 0x2c00 | +| 66 | blk.2.post_ffw_norm_1.weight | 0xc33ca260 | 0x2c00 | +| 67 | blk.2.post_ffw_norm_2.weight | 0xc33cce60 | 0x2c00 | +| 68 | blk.2.pre_ffw_norm_2.weight | 0xc33cfa60 | 0x2c00 | +| 69 | blk.3.attn_k.weight | 0xc33d2660 | 0x5d8000 | +| 70 | blk.3.attn_k_norm.weight | 0xc39aa660 | 0x400 | +| 71 | blk.3.attn_norm.weight | 0xc39aaa60 | 0x2c00 | +| 72 | blk.3.attn_output.weight | 0xc39ad660 | 0xbb0000 | +| 73 | blk.3.attn_q.weight | 0xc455d660 | 0xbb0000 | +| 74 | blk.3.attn_q_norm.weight | 0xc510d660 | 0x400 | +| 75 | blk.3.attn_v.weight | 0xc510da60 | 0x5d8000 | +| 76 | blk.3.ffn_down.weight | 0xc56e5a60 | 0x606c00 | +| 77 | blk.3.ffn_down_exps.scale | 0xc5cec660 | 0x200 | +| 78 | blk.3.ffn_down_exps.weight | 0xc5cec860 | 0x10120000 | +| 79 | blk.3.ffn_gate.weight | 0xd5e0c860 | 0x606c00 | +| 80 | blk.3.ffn_gate_inp.scale | 0xd6413460 | 0x2c00 | +| 81 | blk.3.ffn_gate_inp.weight | 0xd6416060 | 0x160000 | +| 82 | blk.3.ffn_gate_up_exps.weight | 0xd6576060 | 0x20240000 | +| 83 | blk.3.ffn_norm.weight | 0xf67b6060 | 0x2c00 | +| 84 | blk.3.ffn_up.weight | 0xf67b8c60 | 0x606c00 | +| 85 | blk.3.layer_output_scale.weight | 0xf6dbf860 | 0x4 | +| 86 | blk.3.post_attention_norm.weight | 0xf6dbf880 | 0x2c00 | +| 87 | blk.3.post_ffw_norm.weight | 0xf6dc2480 | 0x2c00 | +| 88 | blk.3.post_ffw_norm_1.weight | 0xf6dc5080 | 0x2c00 | +| 89 | blk.3.post_ffw_norm_2.weight | 0xf6dc7c80 | 0x2c00 | +| 90 | blk.3.pre_ffw_norm_2.weight | 0xf6dca880 | 0x2c00 | +| 91 | blk.4.attn_k.weight | 0xf6dcd480 | 0x5d8000 | +| 92 | blk.4.attn_k_norm.weight | 0xf73a5480 | 0x400 | +| 93 | blk.4.attn_norm.weight | 0xf73a5880 | 0x2c00 | +| 94 | blk.4.attn_output.weight | 0xf73a8480 | 0xbb0000 | +| 95 | blk.4.attn_q.weight | 0xf7f58480 | 0xbb0000 | +| 96 | blk.4.attn_q_norm.weight | 0xf8b08480 | 0x400 | +| 97 | blk.4.attn_v.weight | 0xf8b08880 | 0x5d8000 | +| 98 | blk.4.ffn_down.weight | 0xf90e0880 | 0x606c00 | +| 99 | blk.4.ffn_down_exps.scale | 0xf96e7480 | 0x200 | +| 100 | blk.4.ffn_down_exps.weight | 0xf96e7680 | 0x10120000 | +| 101 | blk.4.ffn_gate.weight | 0x109807680 | 0x606c00 | +| 102 | blk.4.ffn_gate_inp.scale | 0x109e0e280 | 0x2c00 | +| 103 | blk.4.ffn_gate_inp.weight | 0x109e10e80 | 0x160000 | +| 104 | blk.4.ffn_gate_up_exps.weight | 0x109f70e80 | 0x20240000 | +| 105 | blk.4.ffn_norm.weight | 0x12a1b0e80 | 0x2c00 | +| 106 | blk.4.ffn_up.weight | 0x12a1b3a80 | 0x606c00 | +| 107 | blk.4.layer_output_scale.weight | 0x12a7ba680 | 0x4 | +| 108 | blk.4.post_attention_norm.weight | 0x12a7ba6a0 | 0x2c00 | +| 109 | blk.4.post_ffw_norm.weight | 0x12a7bd2a0 | 0x2c00 | +| 110 | blk.4.post_ffw_norm_1.weight | 0x12a7bfea0 | 0x2c00 | +| 111 | blk.4.post_ffw_norm_2.weight | 0x12a7c2aa0 | 0x2c00 | +| 112 | blk.4.pre_ffw_norm_2.weight | 0x12a7c56a0 | 0x2c00 | +| 113 | blk.5.attn_k.weight | 0x12a7c82a0 | 0x2ec000 | +| 114 | blk.5.attn_k_norm.weight | 0x12aab42a0 | 0x800 | +| 115 | blk.5.attn_norm.weight | 0x12aab4aa0 | 0x2c00 | +| 116 | blk.5.attn_output.weight | 0x12aab76a0 | 0x1760000 | +| 117 | blk.5.attn_q.weight | 0x12c2176a0 | 0x1760000 | +| 118 | blk.5.attn_q_norm.weight | 0x12d9776a0 | 0x800 | +| 119 | blk.5.ffn_down.weight | 0x12d977ea0 | 0x606c00 | +| 120 | blk.5.ffn_down_exps.scale | 0x12df7eaa0 | 0x200 | +| 121 | blk.5.ffn_down_exps.weight | 0x12df7eca0 | 0x10120000 | +| 122 | blk.5.ffn_gate.weight | 0x13e09eca0 | 0x606c00 | +| 123 | blk.5.ffn_gate_inp.scale | 0x13e6a58a0 | 0x2c00 | +| 124 | blk.5.ffn_gate_inp.weight | 0x13e6a84a0 | 0x160000 | +| 125 | blk.5.ffn_gate_up_exps.weight | 0x13e8084a0 | 0x20240000 | +| 126 | blk.5.ffn_norm.weight | 0x15ea484a0 | 0x2c00 | +| 127 | blk.5.ffn_up.weight | 0x15ea4b0a0 | 0x606c00 | +| 128 | blk.5.layer_output_scale.weight | 0x15f051ca0 | 0x4 | +| 129 | blk.5.post_attention_norm.weight | 0x15f051cc0 | 0x2c00 | +| 130 | blk.5.post_ffw_norm.weight | 0x15f0548c0 | 0x2c00 | +| 131 | blk.5.post_ffw_norm_1.weight | 0x15f0574c0 | 0x2c00 | +| 132 | blk.5.post_ffw_norm_2.weight | 0x15f05a0c0 | 0x2c00 | +| 133 | blk.5.pre_ffw_norm_2.weight | 0x15f05ccc0 | 0x2c00 | +| 134 | blk.6.attn_k.weight | 0x15f05f8c0 | 0x5d8000 | +| 135 | blk.6.attn_k_norm.weight | 0x15f6378c0 | 0x400 | +| 136 | blk.6.attn_norm.weight | 0x15f637cc0 | 0x2c00 | +| 137 | blk.6.attn_output.weight | 0x15f63a8c0 | 0xbb0000 | +| 138 | blk.6.attn_q.weight | 0x1601ea8c0 | 0xbb0000 | +| 139 | blk.6.attn_q_norm.weight | 0x160d9a8c0 | 0x400 | +| 140 | blk.6.attn_v.weight | 0x160d9acc0 | 0x5d8000 | +| 141 | blk.6.ffn_down.weight | 0x161372cc0 | 0x606c00 | +| 142 | blk.6.ffn_down_exps.scale | 0x1619798c0 | 0x200 | +| 143 | blk.6.ffn_down_exps.weight | 0x161979ac0 | 0x10120000 | +| 144 | blk.6.ffn_gate.weight | 0x171a99ac0 | 0x606c00 | +| 145 | blk.6.ffn_gate_inp.scale | 0x1720a06c0 | 0x2c00 | +| 146 | blk.6.ffn_gate_inp.weight | 0x1720a32c0 | 0x160000 | +| 147 | blk.6.ffn_gate_up_exps.weight | 0x1722032c0 | 0x20240000 | +| 148 | blk.6.ffn_norm.weight | 0x1924432c0 | 0x2c00 | +| 149 | blk.6.ffn_up.weight | 0x192445ec0 | 0x606c00 | +| 150 | blk.6.layer_output_scale.weight | 0x192a4cac0 | 0x4 | +| 151 | blk.6.post_attention_norm.weight | 0x192a4cae0 | 0x2c00 | +| 152 | blk.6.post_ffw_norm.weight | 0x192a4f6e0 | 0x2c00 | +| 153 | blk.6.post_ffw_norm_1.weight | 0x192a522e0 | 0x2c00 | +| 154 | blk.6.post_ffw_norm_2.weight | 0x192a54ee0 | 0x2c00 | +| 155 | blk.6.pre_ffw_norm_2.weight | 0x192a57ae0 | 0x2c00 | +| 156 | blk.7.attn_k.weight | 0x192a5a6e0 | 0x5d8000 | +| 157 | blk.7.attn_k_norm.weight | 0x1930326e0 | 0x400 | +| 158 | blk.7.attn_norm.weight | 0x193032ae0 | 0x2c00 | +| 159 | blk.7.attn_output.weight | 0x1930356e0 | 0xbb0000 | +| 160 | blk.7.attn_q.weight | 0x193be56e0 | 0xbb0000 | +| 161 | blk.7.attn_q_norm.weight | 0x1947956e0 | 0x400 | +| 162 | blk.7.attn_v.weight | 0x194795ae0 | 0x5d8000 | +| 163 | blk.7.ffn_down.weight | 0x194d6dae0 | 0x606c00 | +| 164 | blk.7.ffn_down_exps.scale | 0x1953746e0 | 0x200 | +| 165 | blk.7.ffn_down_exps.weight | 0x1953748e0 | 0x10120000 | +| 166 | blk.7.ffn_gate.weight | 0x1a54948e0 | 0x606c00 | +| 167 | blk.7.ffn_gate_inp.scale | 0x1a5a9b4e0 | 0x2c00 | +| 168 | blk.7.ffn_gate_inp.weight | 0x1a5a9e0e0 | 0x160000 | +| 169 | blk.7.ffn_gate_up_exps.weight | 0x1a5bfe0e0 | 0x20240000 | +| 170 | blk.7.ffn_norm.weight | 0x1c5e3e0e0 | 0x2c00 | +| 171 | blk.7.ffn_up.weight | 0x1c5e40ce0 | 0x606c00 | +| 172 | blk.7.layer_output_scale.weight | 0x1c64478e0 | 0x4 | +| 173 | blk.7.post_attention_norm.weight | 0x1c6447900 | 0x2c00 | +| 174 | blk.7.post_ffw_norm.weight | 0x1c644a500 | 0x2c00 | +| 175 | blk.7.post_ffw_norm_1.weight | 0x1c644d100 | 0x2c00 | +| 176 | blk.7.post_ffw_norm_2.weight | 0x1c644fd00 | 0x2c00 | +| 177 | blk.7.pre_ffw_norm_2.weight | 0x1c6452900 | 0x2c00 | +| 178 | blk.8.attn_k.weight | 0x1c6455500 | 0x5d8000 | +| 179 | blk.8.attn_k_norm.weight | 0x1c6a2d500 | 0x400 | +| 180 | blk.8.attn_norm.weight | 0x1c6a2d900 | 0x2c00 | +| 181 | blk.8.attn_output.weight | 0x1c6a30500 | 0xbb0000 | +| 182 | blk.8.attn_q.weight | 0x1c75e0500 | 0xbb0000 | +| 183 | blk.8.attn_q_norm.weight | 0x1c8190500 | 0x400 | +| 184 | blk.8.attn_v.weight | 0x1c8190900 | 0x5d8000 | +| 185 | blk.8.ffn_down.weight | 0x1c8768900 | 0x606c00 | +| 186 | blk.8.ffn_down_exps.scale | 0x1c8d6f500 | 0x200 | +| 187 | blk.8.ffn_down_exps.weight | 0x1c8d6f700 | 0x10120000 | +| 188 | blk.8.ffn_gate.weight | 0x1d8e8f700 | 0x606c00 | +| 189 | blk.8.ffn_gate_inp.scale | 0x1d9496300 | 0x2c00 | +| 190 | blk.8.ffn_gate_inp.weight | 0x1d9498f00 | 0x160000 | +| 191 | blk.8.ffn_gate_up_exps.weight | 0x1d95f8f00 | 0x20240000 | +| 192 | blk.8.ffn_norm.weight | 0x1f9838f00 | 0x2c00 | +| 193 | blk.8.ffn_up.weight | 0x1f983bb00 | 0x606c00 | +| 194 | blk.8.layer_output_scale.weight | 0x1f9e42700 | 0x4 | +| 195 | blk.8.post_attention_norm.weight | 0x1f9e42720 | 0x2c00 | +| 196 | blk.8.post_ffw_norm.weight | 0x1f9e45320 | 0x2c00 | +| 197 | blk.8.post_ffw_norm_1.weight | 0x1f9e47f20 | 0x2c00 | +| 198 | blk.8.post_ffw_norm_2.weight | 0x1f9e4ab20 | 0x2c00 | +| 199 | blk.8.pre_ffw_norm_2.weight | 0x1f9e4d720 | 0x2c00 | +| 200 | blk.9.attn_k.weight | 0x1f9e50320 | 0x5d8000 | +| 201 | blk.9.attn_k_norm.weight | 0x1fa428320 | 0x400 | +| 202 | blk.9.attn_norm.weight | 0x1fa428720 | 0x2c00 | +| 203 | blk.9.attn_output.weight | 0x1fa42b320 | 0xbb0000 | +| 204 | blk.9.attn_q.weight | 0x1fafdb320 | 0xbb0000 | +| 205 | blk.9.attn_q_norm.weight | 0x1fbb8b320 | 0x400 | +| 206 | blk.9.attn_v.weight | 0x1fbb8b720 | 0xb00000 | +| 207 | blk.9.ffn_down.weight | 0x1fc68b720 | 0x606c00 | +| 208 | blk.9.ffn_down_exps.scale | 0x1fcc92320 | 0x200 | +| 209 | blk.9.ffn_down_exps.weight | 0x1fcc92520 | 0x10120000 | +| 210 | blk.9.ffn_gate.weight | 0x20cdb2520 | 0x606c00 | +| 211 | blk.9.ffn_gate_inp.scale | 0x20d3b9120 | 0x2c00 | +| 212 | blk.9.ffn_gate_inp.weight | 0x20d3bbd20 | 0x160000 | +| 213 | blk.9.ffn_gate_up_exps.weight | 0x20d51bd20 | 0x20240000 | +| 214 | blk.9.ffn_norm.weight | 0x22d75bd20 | 0x2c00 | +| 215 | blk.9.ffn_up.weight | 0x22d75e920 | 0x606c00 | +| 216 | blk.9.layer_output_scale.weight | 0x22dd65520 | 0x4 | +| 217 | blk.9.post_attention_norm.weight | 0x22dd65540 | 0x2c00 | +| 218 | blk.9.post_ffw_norm.weight | 0x22dd68140 | 0x2c00 | +| 219 | blk.9.post_ffw_norm_1.weight | 0x22dd6ad40 | 0x2c00 | +| 220 | blk.9.post_ffw_norm_2.weight | 0x22dd6d940 | 0x2c00 | +| 221 | blk.9.pre_ffw_norm_2.weight | 0x22dd70540 | 0x2c00 | +| 222 | blk.10.attn_k.weight | 0x22dd73140 | 0x5d8000 | +| 223 | blk.10.attn_k_norm.weight | 0x22e34b140 | 0x400 | +| 224 | blk.10.attn_norm.weight | 0x22e34b540 | 0x2c00 | +| 225 | blk.10.attn_output.weight | 0x22e34e140 | 0xbb0000 | +| 226 | blk.10.attn_q.weight | 0x22eefe140 | 0xbb0000 | +| 227 | blk.10.attn_q_norm.weight | 0x22faae140 | 0x400 | +| 228 | blk.10.attn_v.weight | 0x22faae540 | 0x5d8000 | +| 229 | blk.10.ffn_down.weight | 0x230086540 | 0x606c00 | +| 230 | blk.10.ffn_down_exps.scale | 0x23068d140 | 0x200 | +| 231 | blk.10.ffn_down_exps.weight | 0x23068d340 | 0x10120000 | +| 232 | blk.10.ffn_gate.weight | 0x2407ad340 | 0x606c00 | +| 233 | blk.10.ffn_gate_inp.scale | 0x240db3f40 | 0x2c00 | +| 234 | blk.10.ffn_gate_inp.weight | 0x240db6b40 | 0x160000 | +| 235 | blk.10.ffn_gate_up_exps.weight | 0x240f16b40 | 0x20240000 | +| 236 | blk.10.ffn_norm.weight | 0x261156b40 | 0x2c00 | +| 237 | blk.10.ffn_up.weight | 0x261159740 | 0x606c00 | +| 238 | blk.10.layer_output_scale.weight | 0x261760340 | 0x4 | +| 239 | blk.10.post_attention_norm.weight | 0x261760360 | 0x2c00 | +| 240 | blk.10.post_ffw_norm.weight | 0x261762f60 | 0x2c00 | +| 241 | blk.10.post_ffw_norm_1.weight | 0x261765b60 | 0x2c00 | +| 242 | blk.10.post_ffw_norm_2.weight | 0x261768760 | 0x2c00 | +| 243 | blk.10.pre_ffw_norm_2.weight | 0x26176b360 | 0x2c00 | +| 244 | blk.11.attn_k.weight | 0x26176df60 | 0x2ec000 | +| 245 | blk.11.attn_k_norm.weight | 0x261a59f60 | 0x800 | +| 246 | blk.11.attn_norm.weight | 0x261a5a760 | 0x2c00 | +| 247 | blk.11.attn_output.weight | 0x261a5d360 | 0x1760000 | +| 248 | blk.11.attn_q.weight | 0x2631bd360 | 0x1760000 | +| 249 | blk.11.attn_q_norm.weight | 0x26491d360 | 0x800 | +| 250 | blk.11.ffn_down.weight | 0x26491db60 | 0xb58000 | +| 251 | blk.11.ffn_down_exps.scale | 0x265475b60 | 0x200 | +| 252 | blk.11.ffn_down_exps.weight | 0x265475d60 | 0x10120000 | +| 253 | blk.11.ffn_gate.weight | 0x275595d60 | 0x606c00 | +| 254 | blk.11.ffn_gate_inp.scale | 0x275b9c960 | 0x2c00 | +| 255 | blk.11.ffn_gate_inp.weight | 0x275b9f560 | 0x160000 | +| 256 | blk.11.ffn_gate_up_exps.weight | 0x275cff560 | 0x20240000 | +| 257 | blk.11.ffn_norm.weight | 0x295f3f560 | 0x2c00 | +| 258 | blk.11.ffn_up.weight | 0x295f42160 | 0x606c00 | +| 259 | blk.11.layer_output_scale.weight | 0x296548d60 | 0x4 | +| 260 | blk.11.post_attention_norm.weight | 0x296548d80 | 0x2c00 | +| 261 | blk.11.post_ffw_norm.weight | 0x29654b980 | 0x2c00 | +| 262 | blk.11.post_ffw_norm_1.weight | 0x29654e580 | 0x2c00 | +| 263 | blk.11.post_ffw_norm_2.weight | 0x296551180 | 0x2c00 | +| 264 | blk.11.pre_ffw_norm_2.weight | 0x296553d80 | 0x2c00 | +| 265 | blk.12.attn_k.weight | 0x296556980 | 0x5d8000 | +| 266 | blk.12.attn_k_norm.weight | 0x296b2e980 | 0x400 | +| 267 | blk.12.attn_norm.weight | 0x296b2ed80 | 0x2c00 | +| 268 | blk.12.attn_output.weight | 0x296b31980 | 0xbb0000 | +| 269 | blk.12.attn_q.weight | 0x2976e1980 | 0xbb0000 | +| 270 | blk.12.attn_q_norm.weight | 0x298291980 | 0x400 | +| 271 | blk.12.attn_v.weight | 0x298291d80 | 0x5d8000 | +| 272 | blk.12.ffn_down.weight | 0x298869d80 | 0x606c00 | +| 273 | blk.12.ffn_down_exps.scale | 0x298e70980 | 0x200 | +| 274 | blk.12.ffn_down_exps.weight | 0x298e70b80 | 0x10120000 | +| 275 | blk.12.ffn_gate.weight | 0x2a8f90b80 | 0x606c00 | +| 276 | blk.12.ffn_gate_inp.scale | 0x2a9597780 | 0x2c00 | +| 277 | blk.12.ffn_gate_inp.weight | 0x2a959a380 | 0x160000 | +| 278 | blk.12.ffn_gate_up_exps.weight | 0x2a96fa380 | 0x20240000 | +| 279 | blk.12.ffn_norm.weight | 0x2c993a380 | 0x2c00 | +| 280 | blk.12.ffn_up.weight | 0x2c993cf80 | 0x606c00 | +| 281 | blk.12.layer_output_scale.weight | 0x2c9f43b80 | 0x4 | +| 282 | blk.12.post_attention_norm.weight | 0x2c9f43ba0 | 0x2c00 | +| 283 | blk.12.post_ffw_norm.weight | 0x2c9f467a0 | 0x2c00 | +| 284 | blk.12.post_ffw_norm_1.weight | 0x2c9f493a0 | 0x2c00 | +| 285 | blk.12.post_ffw_norm_2.weight | 0x2c9f4bfa0 | 0x2c00 | +| 286 | blk.12.pre_ffw_norm_2.weight | 0x2c9f4eba0 | 0x2c00 | +| 287 | blk.13.attn_k.weight | 0x2c9f517a0 | 0x5d8000 | +| 288 | blk.13.attn_k_norm.weight | 0x2ca5297a0 | 0x400 | +| 289 | blk.13.attn_norm.weight | 0x2ca529ba0 | 0x2c00 | +| 290 | blk.13.attn_output.weight | 0x2ca52c7a0 | 0xbb0000 | +| 291 | blk.13.attn_q.weight | 0x2cb0dc7a0 | 0xbb0000 | +| 292 | blk.13.attn_q_norm.weight | 0x2cbc8c7a0 | 0x400 | +| 293 | blk.13.attn_v.weight | 0x2cbc8cba0 | 0x5d8000 | +| 294 | blk.13.ffn_down.weight | 0x2cc264ba0 | 0xb58000 | +| 295 | blk.13.ffn_down_exps.scale | 0x2ccdbcba0 | 0x200 | +| 296 | blk.13.ffn_down_exps.weight | 0x2ccdbcda0 | 0x10120000 | +| 297 | blk.13.ffn_gate.weight | 0x2dcedcda0 | 0x606c00 | +| 298 | blk.13.ffn_gate_inp.scale | 0x2dd4e39a0 | 0x2c00 | +| 299 | blk.13.ffn_gate_inp.weight | 0x2dd4e65a0 | 0x160000 | +| 300 | blk.13.ffn_gate_up_exps.weight | 0x2dd6465a0 | 0x20240000 | +| 301 | blk.13.ffn_norm.weight | 0x2fd8865a0 | 0x2c00 | +| 302 | blk.13.ffn_up.weight | 0x2fd8891a0 | 0x606c00 | +| 303 | blk.13.layer_output_scale.weight | 0x2fde8fda0 | 0x4 | +| 304 | blk.13.post_attention_norm.weight | 0x2fde8fdc0 | 0x2c00 | +| 305 | blk.13.post_ffw_norm.weight | 0x2fde929c0 | 0x2c00 | +| 306 | blk.13.post_ffw_norm_1.weight | 0x2fde955c0 | 0x2c00 | +| 307 | blk.13.post_ffw_norm_2.weight | 0x2fde981c0 | 0x2c00 | +| 308 | blk.13.pre_ffw_norm_2.weight | 0x2fde9adc0 | 0x2c00 | +| 309 | blk.14.attn_k.weight | 0x2fde9d9c0 | 0x5d8000 | +| 310 | blk.14.attn_k_norm.weight | 0x2fe4759c0 | 0x400 | +| 311 | blk.14.attn_norm.weight | 0x2fe475dc0 | 0x2c00 | +| 312 | blk.14.attn_output.weight | 0x2fe4789c0 | 0xbb0000 | +| 313 | blk.14.attn_q.weight | 0x2ff0289c0 | 0xbb0000 | +| 314 | blk.14.attn_q_norm.weight | 0x2ffbd89c0 | 0x400 | +| 315 | blk.14.attn_v.weight | 0x2ffbd8dc0 | 0x5d8000 | +| 316 | blk.14.ffn_down.weight | 0x3001b0dc0 | 0x606c00 | +| 317 | blk.14.ffn_down_exps.scale | 0x3007b79c0 | 0x200 | +| 318 | blk.14.ffn_down_exps.weight | 0x3007b7bc0 | 0x10120000 | +| 319 | blk.14.ffn_gate.weight | 0x3108d7bc0 | 0x606c00 | +| 320 | blk.14.ffn_gate_inp.scale | 0x310ede7c0 | 0x2c00 | +| 321 | blk.14.ffn_gate_inp.weight | 0x310ee13c0 | 0x160000 | +| 322 | blk.14.ffn_gate_up_exps.weight | 0x3110413c0 | 0x20240000 | +| 323 | blk.14.ffn_norm.weight | 0x3312813c0 | 0x2c00 | +| 324 | blk.14.ffn_up.weight | 0x331283fc0 | 0x606c00 | +| 325 | blk.14.layer_output_scale.weight | 0x33188abc0 | 0x4 | +| 326 | blk.14.post_attention_norm.weight | 0x33188abe0 | 0x2c00 | +| 327 | blk.14.post_ffw_norm.weight | 0x33188d7e0 | 0x2c00 | +| 328 | blk.14.post_ffw_norm_1.weight | 0x3318903e0 | 0x2c00 | +| 329 | blk.14.post_ffw_norm_2.weight | 0x331892fe0 | 0x2c00 | +| 330 | blk.14.pre_ffw_norm_2.weight | 0x331895be0 | 0x2c00 | +| 331 | blk.15.attn_k.weight | 0x3318987e0 | 0x5d8000 | +| 332 | blk.15.attn_k_norm.weight | 0x331e707e0 | 0x400 | +| 333 | blk.15.attn_norm.weight | 0x331e70be0 | 0x2c00 | +| 334 | blk.15.attn_output.weight | 0x331e737e0 | 0xbb0000 | +| 335 | blk.15.attn_q.weight | 0x332a237e0 | 0xbb0000 | +| 336 | blk.15.attn_q_norm.weight | 0x3335d37e0 | 0x400 | +| 337 | blk.15.attn_v.weight | 0x3335d3be0 | 0x5d8000 | +| 338 | blk.15.ffn_down.weight | 0x333babbe0 | 0xb58000 | +| 339 | blk.15.ffn_down_exps.scale | 0x334703be0 | 0x200 | +| 340 | blk.15.ffn_down_exps.weight | 0x334703de0 | 0x10120000 | +| 341 | blk.15.ffn_gate.weight | 0x344823de0 | 0x606c00 | +| 342 | blk.15.ffn_gate_inp.scale | 0x344e2a9e0 | 0x2c00 | +| 343 | blk.15.ffn_gate_inp.weight | 0x344e2d5e0 | 0x160000 | +| 344 | blk.15.ffn_gate_up_exps.weight | 0x344f8d5e0 | 0x20240000 | +| 345 | blk.15.ffn_norm.weight | 0x3651cd5e0 | 0x2c00 | +| 346 | blk.15.ffn_up.weight | 0x3651d01e0 | 0x606c00 | +| 347 | blk.15.layer_output_scale.weight | 0x3657d6de0 | 0x4 | +| 348 | blk.15.post_attention_norm.weight | 0x3657d6e00 | 0x2c00 | +| 349 | blk.15.post_ffw_norm.weight | 0x3657d9a00 | 0x2c00 | +| 350 | blk.15.post_ffw_norm_1.weight | 0x3657dc600 | 0x2c00 | +| 351 | blk.15.post_ffw_norm_2.weight | 0x3657df200 | 0x2c00 | +| 352 | blk.15.pre_ffw_norm_2.weight | 0x3657e1e00 | 0x2c00 | +| 353 | blk.16.attn_k.weight | 0x3657e4a00 | 0x5d8000 | +| 354 | blk.16.attn_k_norm.weight | 0x365dbca00 | 0x400 | +| 355 | blk.16.attn_norm.weight | 0x365dbce00 | 0x2c00 | +| 356 | blk.16.attn_output.weight | 0x365dbfa00 | 0xbb0000 | +| 357 | blk.16.attn_q.weight | 0x36696fa00 | 0xbb0000 | +| 358 | blk.16.attn_q_norm.weight | 0x36751fa00 | 0x400 | +| 359 | blk.16.attn_v.weight | 0x36751fe00 | 0x5d8000 | +| 360 | blk.16.ffn_down.weight | 0x367af7e00 | 0xb58000 | +| 361 | blk.16.ffn_down_exps.scale | 0x36864fe00 | 0x200 | +| 362 | blk.16.ffn_down_exps.weight | 0x368650000 | 0x10120000 | +| 363 | blk.16.ffn_gate.weight | 0x378770000 | 0x606c00 | +| 364 | blk.16.ffn_gate_inp.scale | 0x378d76c00 | 0x2c00 | +| 365 | blk.16.ffn_gate_inp.weight | 0x378d79800 | 0x160000 | +| 366 | blk.16.ffn_gate_up_exps.weight | 0x378ed9800 | 0x20240000 | +| 367 | blk.16.ffn_norm.weight | 0x399119800 | 0x2c00 | +| 368 | blk.16.ffn_up.weight | 0x39911c400 | 0x606c00 | +| 369 | blk.16.layer_output_scale.weight | 0x399723000 | 0x4 | +| 370 | blk.16.post_attention_norm.weight | 0x399723020 | 0x2c00 | +| 371 | blk.16.post_ffw_norm.weight | 0x399725c20 | 0x2c00 | +| 372 | blk.16.post_ffw_norm_1.weight | 0x399728820 | 0x2c00 | +| 373 | blk.16.post_ffw_norm_2.weight | 0x39972b420 | 0x2c00 | +| 374 | blk.16.pre_ffw_norm_2.weight | 0x39972e020 | 0x2c00 | +| 375 | blk.17.attn_k.weight | 0x399730c20 | 0x2ec000 | +| 376 | blk.17.attn_k_norm.weight | 0x399a1cc20 | 0x800 | +| 377 | blk.17.attn_norm.weight | 0x399a1d420 | 0x2c00 | +| 378 | blk.17.attn_output.weight | 0x399a20020 | 0x1760000 | +| 379 | blk.17.attn_q.weight | 0x39b180020 | 0x1760000 | +| 380 | blk.17.attn_q_norm.weight | 0x39c8e0020 | 0x800 | +| 381 | blk.17.ffn_down.weight | 0x39c8e0820 | 0x606c00 | +| 382 | blk.17.ffn_down_exps.scale | 0x39cee7420 | 0x200 | +| 383 | blk.17.ffn_down_exps.weight | 0x39cee7620 | 0x10120000 | +| 384 | blk.17.ffn_gate.weight | 0x3ad007620 | 0x606c00 | +| 385 | blk.17.ffn_gate_inp.scale | 0x3ad60e220 | 0x2c00 | +| 386 | blk.17.ffn_gate_inp.weight | 0x3ad610e20 | 0x160000 | +| 387 | blk.17.ffn_gate_up_exps.weight | 0x3ad770e20 | 0x20240000 | +| 388 | blk.17.ffn_norm.weight | 0x3cd9b0e20 | 0x2c00 | +| 389 | blk.17.ffn_up.weight | 0x3cd9b3a20 | 0x606c00 | +| 390 | blk.17.layer_output_scale.weight | 0x3cdfba620 | 0x4 | +| 391 | blk.17.post_attention_norm.weight | 0x3cdfba640 | 0x2c00 | +| 392 | blk.17.post_ffw_norm.weight | 0x3cdfbd240 | 0x2c00 | +| 393 | blk.17.post_ffw_norm_1.weight | 0x3cdfbfe40 | 0x2c00 | +| 394 | blk.17.post_ffw_norm_2.weight | 0x3cdfc2a40 | 0x2c00 | +| 395 | blk.17.pre_ffw_norm_2.weight | 0x3cdfc5640 | 0x2c00 | +| 396 | blk.18.attn_k.weight | 0x3cdfc8240 | 0x5d8000 | +| 397 | blk.18.attn_k_norm.weight | 0x3ce5a0240 | 0x400 | +| 398 | blk.18.attn_norm.weight | 0x3ce5a0640 | 0x2c00 | +| 399 | blk.18.attn_output.weight | 0x3ce5a3240 | 0xbb0000 | +| 400 | blk.18.attn_q.weight | 0x3cf153240 | 0xbb0000 | +| 401 | blk.18.attn_q_norm.weight | 0x3cfd03240 | 0x400 | +| 402 | blk.18.attn_v.weight | 0x3cfd03640 | 0x5d8000 | +| 403 | blk.18.ffn_down.weight | 0x3d02db640 | 0x606c00 | +| 404 | blk.18.ffn_down_exps.scale | 0x3d08e2240 | 0x200 | +| 405 | blk.18.ffn_down_exps.weight | 0x3d08e2440 | 0x10120000 | +| 406 | blk.18.ffn_gate.weight | 0x3e0a02440 | 0x606c00 | +| 407 | blk.18.ffn_gate_inp.scale | 0x3e1009040 | 0x2c00 | +| 408 | blk.18.ffn_gate_inp.weight | 0x3e100bc40 | 0x160000 | +| 409 | blk.18.ffn_gate_up_exps.weight | 0x3e116bc40 | 0x20240000 | +| 410 | blk.18.ffn_norm.weight | 0x4013abc40 | 0x2c00 | +| 411 | blk.18.ffn_up.weight | 0x4013ae840 | 0x606c00 | +| 412 | blk.18.layer_output_scale.weight | 0x4019b5440 | 0x4 | +| 413 | blk.18.post_attention_norm.weight | 0x4019b5460 | 0x2c00 | +| 414 | blk.18.post_ffw_norm.weight | 0x4019b8060 | 0x2c00 | +| 415 | blk.18.post_ffw_norm_1.weight | 0x4019bac60 | 0x2c00 | +| 416 | blk.18.post_ffw_norm_2.weight | 0x4019bd860 | 0x2c00 | +| 417 | blk.18.pre_ffw_norm_2.weight | 0x4019c0460 | 0x2c00 | +| 418 | blk.19.attn_k.weight | 0x4019c3060 | 0x5d8000 | +| 419 | blk.19.attn_k_norm.weight | 0x401f9b060 | 0x400 | +| 420 | blk.19.attn_norm.weight | 0x401f9b460 | 0x2c00 | +| 421 | blk.19.attn_output.weight | 0x401f9e060 | 0xbb0000 | +| 422 | blk.19.attn_q.weight | 0x402b4e060 | 0xbb0000 | +| 423 | blk.19.attn_q_norm.weight | 0x4036fe060 | 0x400 | +| 424 | blk.19.attn_v.weight | 0x4036fe460 | 0xb00000 | +| 425 | blk.19.ffn_down.weight | 0x4041fe460 | 0xb58000 | +| 426 | blk.19.ffn_down_exps.scale | 0x404d56460 | 0x200 | +| 427 | blk.19.ffn_down_exps.weight | 0x404d56660 | 0x10120000 | +| 428 | blk.19.ffn_gate.weight | 0x414e76660 | 0x606c00 | +| 429 | blk.19.ffn_gate_inp.scale | 0x41547d260 | 0x2c00 | +| 430 | blk.19.ffn_gate_inp.weight | 0x41547fe60 | 0x160000 | +| 431 | blk.19.ffn_gate_up_exps.weight | 0x4155dfe60 | 0x20240000 | +| 432 | blk.19.ffn_norm.weight | 0x43581fe60 | 0x2c00 | +| 433 | blk.19.ffn_up.weight | 0x435822a60 | 0x606c00 | +| 434 | blk.19.layer_output_scale.weight | 0x435e29660 | 0x4 | +| 435 | blk.19.post_attention_norm.weight | 0x435e29680 | 0x2c00 | +| 436 | blk.19.post_ffw_norm.weight | 0x435e2c280 | 0x2c00 | +| 437 | blk.19.post_ffw_norm_1.weight | 0x435e2ee80 | 0x2c00 | +| 438 | blk.19.post_ffw_norm_2.weight | 0x435e31a80 | 0x2c00 | +| 439 | blk.19.pre_ffw_norm_2.weight | 0x435e34680 | 0x2c00 | +| 440 | blk.20.attn_k.weight | 0x435e37280 | 0x5d8000 | +| 441 | blk.20.attn_k_norm.weight | 0x43640f280 | 0x400 | +| 442 | blk.20.attn_norm.weight | 0x43640f680 | 0x2c00 | +| 443 | blk.20.attn_output.weight | 0x436412280 | 0xbb0000 | +| 444 | blk.20.attn_q.weight | 0x436fc2280 | 0xbb0000 | +| 445 | blk.20.attn_q_norm.weight | 0x437b72280 | 0x400 | +| 446 | blk.20.attn_v.weight | 0x437b72680 | 0x5d8000 | +| 447 | blk.20.ffn_down.weight | 0x43814a680 | 0xb58000 | +| 448 | blk.20.ffn_down_exps.scale | 0x438ca2680 | 0x200 | +| 449 | blk.20.ffn_down_exps.weight | 0x438ca2880 | 0x10120000 | +| 450 | blk.20.ffn_gate.weight | 0x448dc2880 | 0x606c00 | +| 451 | blk.20.ffn_gate_inp.scale | 0x4493c9480 | 0x2c00 | +| 452 | blk.20.ffn_gate_inp.weight | 0x4493cc080 | 0x160000 | +| 453 | blk.20.ffn_gate_up_exps.weight | 0x44952c080 | 0x20240000 | +| 454 | blk.20.ffn_norm.weight | 0x46976c080 | 0x2c00 | +| 455 | blk.20.ffn_up.weight | 0x46976ec80 | 0x606c00 | +| 456 | blk.20.layer_output_scale.weight | 0x469d75880 | 0x4 | +| 457 | blk.20.post_attention_norm.weight | 0x469d758a0 | 0x2c00 | +| 458 | blk.20.post_ffw_norm.weight | 0x469d784a0 | 0x2c00 | +| 459 | blk.20.post_ffw_norm_1.weight | 0x469d7b0a0 | 0x2c00 | +| 460 | blk.20.post_ffw_norm_2.weight | 0x469d7dca0 | 0x2c00 | +| 461 | blk.20.pre_ffw_norm_2.weight | 0x469d808a0 | 0x2c00 | +| 462 | blk.21.attn_k.weight | 0x469d834a0 | 0x5d8000 | +| 463 | blk.21.attn_k_norm.weight | 0x46a35b4a0 | 0x400 | +| 464 | blk.21.attn_norm.weight | 0x46a35b8a0 | 0x2c00 | +| 465 | blk.21.attn_output.weight | 0x46a35e4a0 | 0xbb0000 | +| 466 | blk.21.attn_q.weight | 0x46af0e4a0 | 0xbb0000 | +| 467 | blk.21.attn_q_norm.weight | 0x46babe4a0 | 0x400 | +| 468 | blk.21.attn_v.weight | 0x46babe8a0 | 0x5d8000 | +| 469 | blk.21.ffn_down.weight | 0x46c0968a0 | 0xb58000 | +| 470 | blk.21.ffn_down_exps.scale | 0x46cbee8a0 | 0x200 | +| 471 | blk.21.ffn_down_exps.weight | 0x46cbeeaa0 | 0x10120000 | +| 472 | blk.21.ffn_gate.weight | 0x47cd0eaa0 | 0x606c00 | +| 473 | blk.21.ffn_gate_inp.scale | 0x47d3156a0 | 0x2c00 | +| 474 | blk.21.ffn_gate_inp.weight | 0x47d3182a0 | 0x160000 | +| 475 | blk.21.ffn_gate_up_exps.weight | 0x47d4782a0 | 0x20240000 | +| 476 | blk.21.ffn_norm.weight | 0x49d6b82a0 | 0x2c00 | +| 477 | blk.21.ffn_up.weight | 0x49d6baea0 | 0x606c00 | +| 478 | blk.21.layer_output_scale.weight | 0x49dcc1aa0 | 0x4 | +| 479 | blk.21.post_attention_norm.weight | 0x49dcc1ac0 | 0x2c00 | +| 480 | blk.21.post_ffw_norm.weight | 0x49dcc46c0 | 0x2c00 | +| 481 | blk.21.post_ffw_norm_1.weight | 0x49dcc72c0 | 0x2c00 | +| 482 | blk.21.post_ffw_norm_2.weight | 0x49dcc9ec0 | 0x2c00 | +| 483 | blk.21.pre_ffw_norm_2.weight | 0x49dcccac0 | 0x2c00 | +| 484 | blk.22.attn_k.weight | 0x49dccf6c0 | 0x5d8000 | +| 485 | blk.22.attn_k_norm.weight | 0x49e2a76c0 | 0x400 | +| 486 | blk.22.attn_norm.weight | 0x49e2a7ac0 | 0x2c00 | +| 487 | blk.22.attn_output.weight | 0x49e2aa6c0 | 0xbb0000 | +| 488 | blk.22.attn_q.weight | 0x49ee5a6c0 | 0xbb0000 | +| 489 | blk.22.attn_q_norm.weight | 0x49fa0a6c0 | 0x400 | +| 490 | blk.22.attn_v.weight | 0x49fa0aac0 | 0x5d8000 | +| 491 | blk.22.ffn_down.weight | 0x49ffe2ac0 | 0x606c00 | +| 492 | blk.22.ffn_down_exps.scale | 0x4a05e96c0 | 0x200 | +| 493 | blk.22.ffn_down_exps.weight | 0x4a05e98c0 | 0x10120000 | +| 494 | blk.22.ffn_gate.weight | 0x4b07098c0 | 0x606c00 | +| 495 | blk.22.ffn_gate_inp.scale | 0x4b0d104c0 | 0x2c00 | +| 496 | blk.22.ffn_gate_inp.weight | 0x4b0d130c0 | 0x160000 | +| 497 | blk.22.ffn_gate_up_exps.weight | 0x4b0e730c0 | 0x20240000 | +| 498 | blk.22.ffn_norm.weight | 0x4d10b30c0 | 0x2c00 | +| 499 | blk.22.ffn_up.weight | 0x4d10b5cc0 | 0x606c00 | +| 500 | blk.22.layer_output_scale.weight | 0x4d16bc8c0 | 0x4 | +| 501 | blk.22.post_attention_norm.weight | 0x4d16bc8e0 | 0x2c00 | +| 502 | blk.22.post_ffw_norm.weight | 0x4d16bf4e0 | 0x2c00 | +| 503 | blk.22.post_ffw_norm_1.weight | 0x4d16c20e0 | 0x2c00 | +| 504 | blk.22.post_ffw_norm_2.weight | 0x4d16c4ce0 | 0x2c00 | +| 505 | blk.22.pre_ffw_norm_2.weight | 0x4d16c78e0 | 0x2c00 | +| 506 | blk.23.attn_k.weight | 0x4d16ca4e0 | 0x2ec000 | +| 507 | blk.23.attn_k_norm.weight | 0x4d19b64e0 | 0x800 | +| 508 | blk.23.attn_norm.weight | 0x4d19b6ce0 | 0x2c00 | +| 509 | blk.23.attn_output.weight | 0x4d19b98e0 | 0x1760000 | +| 510 | blk.23.attn_q.weight | 0x4d31198e0 | 0x1760000 | +| 511 | blk.23.attn_q_norm.weight | 0x4d48798e0 | 0x800 | +| 512 | blk.23.ffn_down.weight | 0x4d487a0e0 | 0x606c00 | +| 513 | blk.23.ffn_down_exps.scale | 0x4d4e80ce0 | 0x200 | +| 514 | blk.23.ffn_down_exps.weight | 0x4d4e80ee0 | 0x10120000 | +| 515 | blk.23.ffn_gate.weight | 0x4e4fa0ee0 | 0x606c00 | +| 516 | blk.23.ffn_gate_inp.scale | 0x4e55a7ae0 | 0x2c00 | +| 517 | blk.23.ffn_gate_inp.weight | 0x4e55aa6e0 | 0x160000 | +| 518 | blk.23.ffn_gate_up_exps.weight | 0x4e570a6e0 | 0x20240000 | +| 519 | blk.23.ffn_norm.weight | 0x50594a6e0 | 0x2c00 | +| 520 | blk.23.ffn_up.weight | 0x50594d2e0 | 0x606c00 | +| 521 | blk.23.layer_output_scale.weight | 0x505f53ee0 | 0x4 | +| 522 | blk.23.post_attention_norm.weight | 0x505f53f00 | 0x2c00 | +| 523 | blk.23.post_ffw_norm.weight | 0x505f56b00 | 0x2c00 | +| 524 | blk.23.post_ffw_norm_1.weight | 0x505f59700 | 0x2c00 | +| 525 | blk.23.post_ffw_norm_2.weight | 0x505f5c300 | 0x2c00 | +| 526 | blk.23.pre_ffw_norm_2.weight | 0x505f5ef00 | 0x2c00 | +| 527 | blk.24.attn_k.weight | 0x505f61b00 | 0x5d8000 | +| 528 | blk.24.attn_k_norm.weight | 0x506539b00 | 0x400 | +| 529 | blk.24.attn_norm.weight | 0x506539f00 | 0x2c00 | +| 530 | blk.24.attn_output.weight | 0x50653cb00 | 0xbb0000 | +| 531 | blk.24.attn_q.weight | 0x5070ecb00 | 0xbb0000 | +| 532 | blk.24.attn_q_norm.weight | 0x507c9cb00 | 0x400 | +| 533 | blk.24.attn_v.weight | 0x507c9cf00 | 0x5d8000 | +| 534 | blk.24.ffn_down.weight | 0x508274f00 | 0xb58000 | +| 535 | blk.24.ffn_down_exps.scale | 0x508dccf00 | 0x200 | +| 536 | blk.24.ffn_down_exps.weight | 0x508dcd100 | 0x10120000 | +| 537 | blk.24.ffn_gate.weight | 0x518eed100 | 0x606c00 | +| 538 | blk.24.ffn_gate_inp.scale | 0x5194f3d00 | 0x2c00 | +| 539 | blk.24.ffn_gate_inp.weight | 0x5194f6900 | 0x160000 | +| 540 | blk.24.ffn_gate_up_exps.weight | 0x519656900 | 0x20240000 | +| 541 | blk.24.ffn_norm.weight | 0x539896900 | 0x2c00 | +| 542 | blk.24.ffn_up.weight | 0x539899500 | 0x606c00 | +| 543 | blk.24.layer_output_scale.weight | 0x539ea0100 | 0x4 | +| 544 | blk.24.post_attention_norm.weight | 0x539ea0120 | 0x2c00 | +| 545 | blk.24.post_ffw_norm.weight | 0x539ea2d20 | 0x2c00 | +| 546 | blk.24.post_ffw_norm_1.weight | 0x539ea5920 | 0x2c00 | +| 547 | blk.24.post_ffw_norm_2.weight | 0x539ea8520 | 0x2c00 | +| 548 | blk.24.pre_ffw_norm_2.weight | 0x539eab120 | 0x2c00 | +| 549 | blk.25.attn_k.weight | 0x539eadd20 | 0x5d8000 | +| 550 | blk.25.attn_k_norm.weight | 0x53a485d20 | 0x400 | +| 551 | blk.25.attn_norm.weight | 0x53a486120 | 0x2c00 | +| 552 | blk.25.attn_output.weight | 0x53a488d20 | 0x1600000 | +| 553 | blk.25.attn_q.weight | 0x53ba88d20 | 0xbb0000 | +| 554 | blk.25.attn_q_norm.weight | 0x53c638d20 | 0x400 | +| 555 | blk.25.attn_v.weight | 0x53c639120 | 0x5d8000 | +| 556 | blk.25.ffn_down.weight | 0x53cc11120 | 0x606c00 | +| 557 | blk.25.ffn_down_exps.scale | 0x53d217d20 | 0x200 | +| 558 | blk.25.ffn_down_exps.weight | 0x53d217f20 | 0x10120000 | +| 559 | blk.25.ffn_gate.weight | 0x54d337f20 | 0x606c00 | +| 560 | blk.25.ffn_gate_inp.scale | 0x54d93eb20 | 0x2c00 | +| 561 | blk.25.ffn_gate_inp.weight | 0x54d941720 | 0x160000 | +| 562 | blk.25.ffn_gate_up_exps.weight | 0x54daa1720 | 0x20240000 | +| 563 | blk.25.ffn_norm.weight | 0x56dce1720 | 0x2c00 | +| 564 | blk.25.ffn_up.weight | 0x56dce4320 | 0x606c00 | +| 565 | blk.25.layer_output_scale.weight | 0x56e2eaf20 | 0x4 | +| 566 | blk.25.post_attention_norm.weight | 0x56e2eaf40 | 0x2c00 | +| 567 | blk.25.post_ffw_norm.weight | 0x56e2edb40 | 0x2c00 | +| 568 | blk.25.post_ffw_norm_1.weight | 0x56e2f0740 | 0x2c00 | +| 569 | blk.25.post_ffw_norm_2.weight | 0x56e2f3340 | 0x2c00 | +| 570 | blk.25.pre_ffw_norm_2.weight | 0x56e2f5f40 | 0x2c00 | +| 571 | blk.26.attn_k.weight | 0x56e2f8b40 | 0x5d8000 | +| 572 | blk.26.attn_k_norm.weight | 0x56e8d0b40 | 0x400 | +| 573 | blk.26.attn_norm.weight | 0x56e8d0f40 | 0x2c00 | +| 574 | blk.26.attn_output.weight | 0x56e8d3b40 | 0xbb0000 | +| 575 | blk.26.attn_q.weight | 0x56f483b40 | 0xbb0000 | +| 576 | blk.26.attn_q_norm.weight | 0x570033b40 | 0x400 | +| 577 | blk.26.attn_v.weight | 0x570033f40 | 0xb00000 | +| 578 | blk.26.ffn_down.weight | 0x570b33f40 | 0x606c00 | +| 579 | blk.26.ffn_down_exps.scale | 0x57113ab40 | 0x200 | +| 580 | blk.26.ffn_down_exps.weight | 0x57113ad40 | 0x10120000 | +| 581 | blk.26.ffn_gate.weight | 0x58125ad40 | 0x606c00 | +| 582 | blk.26.ffn_gate_inp.scale | 0x581861940 | 0x2c00 | +| 583 | blk.26.ffn_gate_inp.weight | 0x581864540 | 0x160000 | +| 584 | blk.26.ffn_gate_up_exps.weight | 0x5819c4540 | 0x20240000 | +| 585 | blk.26.ffn_norm.weight | 0x5a1c04540 | 0x2c00 | +| 586 | blk.26.ffn_up.weight | 0x5a1c07140 | 0x606c00 | +| 587 | blk.26.layer_output_scale.weight | 0x5a220dd40 | 0x4 | +| 588 | blk.26.post_attention_norm.weight | 0x5a220dd60 | 0x2c00 | +| 589 | blk.26.post_ffw_norm.weight | 0x5a2210960 | 0x2c00 | +| 590 | blk.26.post_ffw_norm_1.weight | 0x5a2213560 | 0x2c00 | +| 591 | blk.26.post_ffw_norm_2.weight | 0x5a2216160 | 0x2c00 | +| 592 | blk.26.pre_ffw_norm_2.weight | 0x5a2218d60 | 0x2c00 | +| 593 | blk.27.attn_k.weight | 0x5a221b960 | 0x5d8000 | +| 594 | blk.27.attn_k_norm.weight | 0x5a27f3960 | 0x400 | +| 595 | blk.27.attn_norm.weight | 0x5a27f3d60 | 0x2c00 | +| 596 | blk.27.attn_output.weight | 0x5a27f6960 | 0xbb0000 | +| 597 | blk.27.attn_q.weight | 0x5a33a6960 | 0xbb0000 | +| 598 | blk.27.attn_q_norm.weight | 0x5a3f56960 | 0x400 | +| 599 | blk.27.attn_v.weight | 0x5a3f56d60 | 0x5d8000 | +| 600 | blk.27.ffn_down.weight | 0x5a452ed60 | 0x606c00 | +| 601 | blk.27.ffn_down_exps.scale | 0x5a4b35960 | 0x200 | +| 602 | blk.27.ffn_down_exps.weight | 0x5a4b35b60 | 0x10120000 | +| 603 | blk.27.ffn_gate.weight | 0x5b4c55b60 | 0x606c00 | +| 604 | blk.27.ffn_gate_inp.scale | 0x5b525c760 | 0x2c00 | +| 605 | blk.27.ffn_gate_inp.weight | 0x5b525f360 | 0x160000 | +| 606 | blk.27.ffn_gate_up_exps.weight | 0x5b53bf360 | 0x20240000 | +| 607 | blk.27.ffn_norm.weight | 0x5d55ff360 | 0x2c00 | +| 608 | blk.27.ffn_up.weight | 0x5d5601f60 | 0x606c00 | +| 609 | blk.27.layer_output_scale.weight | 0x5d5c08b60 | 0x4 | +| 610 | blk.27.post_attention_norm.weight | 0x5d5c08b80 | 0x2c00 | +| 611 | blk.27.post_ffw_norm.weight | 0x5d5c0b780 | 0x2c00 | +| 612 | blk.27.post_ffw_norm_1.weight | 0x5d5c0e380 | 0x2c00 | +| 613 | blk.27.post_ffw_norm_2.weight | 0x5d5c10f80 | 0x2c00 | +| 614 | blk.27.pre_ffw_norm_2.weight | 0x5d5c13b80 | 0x2c00 | +| 615 | blk.28.attn_k.weight | 0x5d5c16780 | 0x5d8000 | +| 616 | blk.28.attn_k_norm.weight | 0x5d61ee780 | 0x400 | +| 617 | blk.28.attn_norm.weight | 0x5d61eeb80 | 0x2c00 | +| 618 | blk.28.attn_output.weight | 0x5d61f1780 | 0x1600000 | +| 619 | blk.28.attn_q.weight | 0x5d77f1780 | 0xbb0000 | +| 620 | blk.28.attn_q_norm.weight | 0x5d83a1780 | 0x400 | +| 621 | blk.28.attn_v.weight | 0x5d83a1b80 | 0xb00000 | +| 622 | blk.28.ffn_down.weight | 0x5d8ea1b80 | 0x606c00 | +| 623 | blk.28.ffn_down_exps.scale | 0x5d94a8780 | 0x200 | +| 624 | blk.28.ffn_down_exps.weight | 0x5d94a8980 | 0x10120000 | +| 625 | blk.28.ffn_gate.weight | 0x5e95c8980 | 0x606c00 | +| 626 | blk.28.ffn_gate_inp.scale | 0x5e9bcf580 | 0x2c00 | +| 627 | blk.28.ffn_gate_inp.weight | 0x5e9bd2180 | 0x160000 | +| 628 | blk.28.ffn_gate_up_exps.weight | 0x5e9d32180 | 0x20240000 | +| 629 | blk.28.ffn_norm.weight | 0x609f72180 | 0x2c00 | +| 630 | blk.28.ffn_up.weight | 0x609f74d80 | 0x606c00 | +| 631 | blk.28.layer_output_scale.weight | 0x60a57b980 | 0x4 | +| 632 | blk.28.post_attention_norm.weight | 0x60a57b9a0 | 0x2c00 | +| 633 | blk.28.post_ffw_norm.weight | 0x60a57e5a0 | 0x2c00 | +| 634 | blk.28.post_ffw_norm_1.weight | 0x60a5811a0 | 0x2c00 | +| 635 | blk.28.post_ffw_norm_2.weight | 0x60a583da0 | 0x2c00 | +| 636 | blk.28.pre_ffw_norm_2.weight | 0x60a5869a0 | 0x2c00 | +| 637 | blk.29.attn_k.weight | 0x60a5895a0 | 0x2ec000 | +| 638 | blk.29.attn_k_norm.weight | 0x60a8755a0 | 0x800 | +| 639 | blk.29.attn_norm.weight | 0x60a875da0 | 0x2c00 | +| 640 | blk.29.attn_output.weight | 0x60a8789a0 | 0x1760000 | +| 641 | blk.29.attn_q.weight | 0x60bfd89a0 | 0x1760000 | +| 642 | blk.29.attn_q_norm.weight | 0x60d7389a0 | 0x800 | +| 643 | blk.29.ffn_down.weight | 0x60d7391a0 | 0x606c00 | +| 644 | blk.29.ffn_down_exps.scale | 0x60dd3fda0 | 0x200 | +| 645 | blk.29.ffn_down_exps.weight | 0x60dd3ffa0 | 0x10120000 | +| 646 | blk.29.ffn_gate.weight | 0x61de5ffa0 | 0x606c00 | +| 647 | blk.29.ffn_gate_inp.scale | 0x61e466ba0 | 0x2c00 | +| 648 | blk.29.ffn_gate_inp.weight | 0x61e4697a0 | 0x160000 | +| 649 | blk.29.ffn_gate_up_exps.weight | 0x61e5c97a0 | 0x20240000 | +| 650 | blk.29.ffn_norm.weight | 0x63e8097a0 | 0x2c00 | +| 651 | blk.29.ffn_up.weight | 0x63e80c3a0 | 0x606c00 | +| 652 | blk.29.layer_output_scale.weight | 0x63ee12fa0 | 0x4 | +| 653 | blk.29.post_attention_norm.weight | 0x63ee12fc0 | 0x2c00 | +| 654 | blk.29.post_ffw_norm.weight | 0x63ee15bc0 | 0x2c00 | +| 655 | blk.29.post_ffw_norm_1.weight | 0x63ee187c0 | 0x2c00 | +| 656 | blk.29.post_ffw_norm_2.weight | 0x63ee1b3c0 | 0x2c00 | +| 657 | blk.29.pre_ffw_norm_2.weight | 0x63ee1dfc0 | 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 | Q8_0 | 8.5000 | + +- Total elements in base: (~738M) 738200576 +- Percentage of total elements: 2.93% +- Bits per Weight (BPW) for base: 8.5001 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 7 | blk.0.attn_q.weight | Block 0 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 10 | blk.0.ffn_down.weight | Block 0 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 13 | blk.0.ffn_gate.weight | Block 0 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q6_K | 6.5625 | +| 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 | Q8_0 | 8.5000 | +| 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: 7.3032 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 29 | blk.1.attn_q.weight | Block 1 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 32 | blk.1.ffn_down.weight | Block 1 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 35 | blk.1.ffn_gate.weight | Block 1 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 51 | blk.2.attn_q.weight | Block 2 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 54 | blk.2.ffn_down.weight | Block 2 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 57 | blk.2.ffn_gate.weight | Block 2 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 73 | blk.3.attn_q.weight | Block 3 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 76 | blk.3.ffn_down.weight | Block 3 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 79 | blk.3.ffn_gate.weight | Block 3 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 95 | blk.4.attn_q.weight | Block 4 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 98 | blk.4.ffn_down.weight | Block 4 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 101 | blk.4.ffn_gate.weight | Block 4 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 117 | blk.5.attn_q.weight | Block 5 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 122 | blk.5.ffn_gate.weight | Block 5 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5109 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 138 | blk.6.attn_q.weight | Block 6 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 141 | blk.6.ffn_down.weight | Block 6 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 144 | blk.6.ffn_gate.weight | Block 6 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 160 | blk.7.attn_q.weight | Block 7 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 163 | blk.7.ffn_down.weight | Block 7 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 166 | blk.7.ffn_gate.weight | Block 7 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 182 | blk.8.attn_q.weight | Block 8 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 185 | blk.8.ffn_down.weight | Block 8 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 188 | blk.8.ffn_gate.weight | Block 8 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 204 | blk.9.attn_q.weight | Block 9 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 207 | blk.9.ffn_down.weight | Block 9 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 210 | blk.9.ffn_gate.weight | Block 9 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5642 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 226 | blk.10.attn_q.weight | Block 10 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 229 | blk.10.ffn_down.weight | Block 10 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 232 | blk.10.ffn_gate.weight | Block 10 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 248 | blk.11.attn_q.weight | Block 11 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 253 | blk.11.ffn_gate.weight | Block 11 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5647 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 269 | blk.12.attn_q.weight | Block 12 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 272 | blk.12.ffn_down.weight | Block 12 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 275 | blk.12.ffn_gate.weight | Block 12 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 291 | blk.13.attn_q.weight | Block 13 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 294 | blk.13.ffn_down.weight | Block 13 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 297 | blk.13.ffn_gate.weight | Block 13 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5659 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 313 | blk.14.attn_q.weight | Block 14 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 316 | blk.14.ffn_down.weight | Block 14 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 319 | blk.14.ffn_gate.weight | Block 14 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 335 | blk.15.attn_q.weight | Block 15 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 338 | blk.15.ffn_down.weight | Block 15 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 341 | blk.15.ffn_gate.weight | Block 15 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5659 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 357 | blk.16.attn_q.weight | Block 16 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 360 | blk.16.ffn_down.weight | Block 16 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 363 | blk.16.ffn_gate.weight | Block 16 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5659 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 379 | blk.17.attn_q.weight | Block 17 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 384 | blk.17.ffn_gate.weight | Block 17 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5109 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 400 | blk.18.attn_q.weight | Block 18 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 403 | blk.18.ffn_down.weight | Block 18 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 406 | blk.18.ffn_gate.weight | Block 18 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 422 | blk.19.attn_q.weight | Block 19 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 425 | blk.19.ffn_down.weight | Block 19 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 428 | blk.19.ffn_gate.weight | Block 19 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.6190 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 444 | blk.20.attn_q.weight | Block 20 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 447 | blk.20.ffn_down.weight | Block 20 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 450 | blk.20.ffn_gate.weight | Block 20 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5659 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 466 | blk.21.attn_q.weight | Block 21 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 469 | blk.21.ffn_down.weight | Block 21 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 472 | blk.21.ffn_gate.weight | Block 21 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5659 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 488 | blk.22.attn_q.weight | Block 22 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 491 | blk.22.ffn_down.weight | Block 22 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 494 | blk.22.ffn_gate.weight | Block 22 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 510 | blk.23.attn_q.weight | Block 23 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 515 | blk.23.ffn_gate.weight | Block 23 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5109 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 531 | blk.24.attn_q.weight | Block 24 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 534 | blk.24.ffn_down.weight | Block 24 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | F16 | 16.0000 | +| 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 | Q8_0 | 8.5000 | +| 537 | blk.24.ffn_gate.weight | Block 24 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5659 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 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 553 | blk.25.attn_q.weight | Block 25 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 556 | blk.25.ffn_down.weight | Block 25 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 559 | blk.25.ffn_gate.weight | Block 25 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.6173 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 575 | blk.26.attn_q.weight | Block 26 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 578 | blk.26.ffn_down.weight | Block 26 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 581 | blk.26.ffn_gate.weight | Block 26 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5642 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 597 | blk.27.attn_q.weight | Block 27 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 600 | blk.27.ffn_down.weight | Block 27 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 603 | blk.27.ffn_gate.weight | Block 27 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5111 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 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 619 | blk.28.attn_q.weight | Block 28 Attention Query (W) | ( ~12M) 11534336 | 2816 x 4096 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | F16 | 16.0000 | +| 622 | blk.28.ffn_down.weight | Block 28 Feed-Forward Network "Down" (W) | ( ~6M) 5947392 | 2112 x 2816 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 625 | blk.28.ffn_gate.weight | Block 28 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.6705 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 641 | blk.29.attn_q.weight | Block 29 Attention Query (W) | ( ~23M) 23068672 | 2816 x 8192 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 646 | blk.29.ffn_gate.weight | Block 29 Feed-Forward Network "Gate" (W) | ( ~6M) 5947392 | 2816 x 2112 x 1 x 1 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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 | Q8_0 | 8.5000 | +| 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: 8.5109 bits + + +Total BPW for gemma-4-26B-A4B-it-Q8_0.gguf: 8.4996 bits