| { | |
| "metadata": { | |
| "ParamSize": 194, | |
| "ParamBytes": 1447284480.0, | |
| "BitsPerParam": 32.0 | |
| }, | |
| "records": [ | |
| { | |
| "dataPath": "params_shard_0.bin", | |
| "format": "raw-shard", | |
| "nbytes": 94371840, | |
| "records": [ | |
| { | |
| "name": "model.embed_tokens.weight", | |
| "shape": [ | |
| 49152, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 94371840, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "22da8ab49794d5908bc01239610bd5e3" | |
| }, | |
| { | |
| "dataPath": "params_shard_1.bin", | |
| "format": "raw-shard", | |
| "nbytes": 24581760, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1920 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 4917120 | |
| }, | |
| { | |
| "name": "model.layers.0.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 17821440 | |
| }, | |
| { | |
| "name": "model.layers.1.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 19666560 | |
| } | |
| ], | |
| "md5sum": "3455513031408c2af861eba70f886144" | |
| }, | |
| { | |
| "dataPath": "params_shard_2.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.1.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.10.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.10.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.10.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.10.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.10.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "397cafc55ca40192036183358d099274" | |
| }, | |
| { | |
| "dataPath": "params_shard_3.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.11.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.11.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.11.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.11.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.12.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.12.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "efbb69b92372f6951c3cc800dc7dc10b" | |
| }, | |
| { | |
| "dataPath": "params_shard_4.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.12.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.12.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.13.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.13.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.13.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.13.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.13.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "9088e6bd76c248e03d421bf018f75efa" | |
| }, | |
| { | |
| "dataPath": "params_shard_5.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.13.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.14.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.14.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.14.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.14.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.15.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.15.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "7a69ef249750a9f069bfcf4b9be03c2e" | |
| }, | |
| { | |
| "dataPath": "params_shard_6.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.15.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.15.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.16.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.16.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.16.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.16.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.16.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "764e0cf62c2966328c0eb4f3e8c864fa" | |
| }, | |
| { | |
| "dataPath": "params_shard_7.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.17.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.17.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.17.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.17.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.18.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.18.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "af08efb2d0c28c46ec91467dd68ac361" | |
| }, | |
| { | |
| "dataPath": "params_shard_8.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.18.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.18.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.19.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.19.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.19.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.19.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.19.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "1f17fcc52b5f5daa546000fde63cddbe" | |
| }, | |
| { | |
| "dataPath": "params_shard_9.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.19.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.2.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.2.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.2.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.2.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.20.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.20.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "36171e79d7e3df56ff92903f1cd3bea3" | |
| }, | |
| { | |
| "dataPath": "params_shard_10.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.20.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.20.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.21.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.21.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.21.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.21.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.21.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "ef7a6f039eabff1178996715ac8eeda0" | |
| }, | |
| { | |
| "dataPath": "params_shard_11.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.22.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.22.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.22.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.22.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.23.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.23.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "d9a649c7a7fbef5edc5c9d331a9504e0" | |
| }, | |
| { | |
| "dataPath": "params_shard_12.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.23.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.23.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.24.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.24.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.24.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.24.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.24.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "12fd2a3cba331815232faa25e2f76bbd" | |
| }, | |
| { | |
| "dataPath": "params_shard_13.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.25.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.25.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.25.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.25.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.25.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.25.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.26.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.26.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "8d5e9487ffb9775b7a14cf26e27bd251" | |
| }, | |
| { | |
| "dataPath": "params_shard_14.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.26.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.26.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.26.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.26.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.27.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.27.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.27.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.27.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.27.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "fe80334f71e180c122c4cc24c6675835" | |
| }, | |
| { | |
| "dataPath": "params_shard_15.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.27.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.28.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.28.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.28.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.28.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.28.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.28.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.29.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.29.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "9dbb293e707b26ce5256589b533dd335" | |
| }, | |
| { | |
| "dataPath": "params_shard_16.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.29.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.29.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.29.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.29.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.3.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.3.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "ae7c4f24535aef8841ff0e6e2615872b" | |
| }, | |
| { | |
| "dataPath": "params_shard_17.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.30.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.30.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.30.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.30.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.30.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.30.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.31.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.31.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "fb992b59ad6b0aff109e2027167400bd" | |
| }, | |
| { | |
| "dataPath": "params_shard_18.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.31.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.31.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.31.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.31.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.4.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.4.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.4.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.4.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "617763ed09849f428dc86a125311b4fc" | |
| }, | |
| { | |
| "dataPath": "params_shard_19.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.5.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.5.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.6.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.6.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "6a7fbc46c6f54ff39c616c1deef3c4aa" | |
| }, | |
| { | |
| "dataPath": "params_shard_20.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32568960, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.6.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.6.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.6.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.layers.7.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| }, | |
| { | |
| "name": "model.layers.7.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 14749440 | |
| }, | |
| { | |
| "name": "model.layers.7.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.7.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 29495040 | |
| }, | |
| { | |
| "name": "model.layers.7.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 29496960 | |
| } | |
| ], | |
| "md5sum": "ede385b1568d001a41ab667e9f527c4f" | |
| }, | |
| { | |
| "dataPath": "params_shard_21.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26424960, | |
| "records": [ | |
| { | |
| "name": "model.layers.7.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.8.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 1843200 | |
| }, | |
| { | |
| "name": "model.layers.8.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 1845120 | |
| }, | |
| { | |
| "name": "model.layers.8.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 6760320 | |
| }, | |
| { | |
| "name": "model.layers.8.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 16590720 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 16592640 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 19664640 | |
| }, | |
| { | |
| "name": "model.layers.9.input_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 21507840 | |
| }, | |
| { | |
| "name": "model.layers.9.mlp.down_proj.weight", | |
| "shape": [ | |
| 960, | |
| 2560 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4915200, | |
| "byteOffset": 21509760 | |
| } | |
| ], | |
| "md5sum": "db03c78ccc57f144e6b1b2ca25c70e08" | |
| }, | |
| { | |
| "dataPath": "params_shard_22.bin", | |
| "format": "raw-shard", | |
| "nbytes": 14749440, | |
| "records": [ | |
| { | |
| "name": "model.layers.9.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 5120, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 9830400, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.9.post_attention_layernorm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 9830400 | |
| }, | |
| { | |
| "name": "model.layers.9.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 1600, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3072000, | |
| "byteOffset": 9832320 | |
| }, | |
| { | |
| "name": "model.layers.9.self_attn.o_proj.weight", | |
| "shape": [ | |
| 960, | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1843200, | |
| "byteOffset": 12904320 | |
| }, | |
| { | |
| "name": "model.norm.weight", | |
| "shape": [ | |
| 960 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1920, | |
| "byteOffset": 14747520 | |
| } | |
| ], | |
| "md5sum": "4a0b8950406ba4ce5b50c109fdcb5283" | |
| } | |
| ] | |
| } |