| { | |
| "metadata": { | |
| "ParamSize": 65, | |
| "ParamBytes": 830578688.0, | |
| "BitsPerParam": 4.500414746671623 | |
| }, | |
| "records": [ | |
| { | |
| "dataPath": "params_shard_0.bin", | |
| "format": "raw-shard", | |
| "nbytes": 65536000, | |
| "records": [ | |
| { | |
| "name": "model.embed_tokens.q_weight", | |
| "shape": [ | |
| 32000, | |
| 512 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 65536000, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "34491ecb22b5e7f4ca56d82527c438f9" | |
| }, | |
| { | |
| "dataPath": "params_shard_1.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33357824, | |
| "records": [ | |
| { | |
| "name": "model.embed_tokens.q_scale", | |
| "shape": [ | |
| 32000, | |
| 128 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192000, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 12288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 25165824, | |
| "byteOffset": 8192000 | |
| } | |
| ], | |
| "md5sum": "661089d83913d30e7575835461db0abf" | |
| }, | |
| { | |
| "dataPath": "params_shard_2.bin", | |
| "format": "raw-shard", | |
| "nbytes": 45088768, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 22016 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 45088768, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "78b992a9d971e604e70b413c233a60c8" | |
| }, | |
| { | |
| "dataPath": "params_shard_3.bin", | |
| "format": "raw-shard", | |
| "nbytes": 22544384, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 1376, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 22544384, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2120fd58f3fa655d99cc9fe71bfbee19" | |
| }, | |
| { | |
| "dataPath": "params_shard_4.bin", | |
| "format": "raw-shard", | |
| "nbytes": 25165824, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 12288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 25165824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c15d92613464fb1db85970571ee48e90" | |
| }, | |
| { | |
| "dataPath": "params_shard_5.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32587776, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 12288 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3145728, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8388608, | |
| "byteOffset": 3145728 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1048576, | |
| "byteOffset": 11534336 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 22016 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 5636096, | |
| "byteOffset": 12582912 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 344, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2818048, | |
| "byteOffset": 18219008 | |
| }, | |
| { | |
| "name": "model.layers.0.input_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 21037056 | |
| }, | |
| { | |
| "name": "model.layers.0.post_attention_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 21045248 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 12288 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3145728, | |
| "byteOffset": 21053440 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8388608, | |
| "byteOffset": 24199168 | |
| } | |
| ], | |
| "md5sum": "dfe551efa483fbf60228e22987a26a32" | |
| }, | |
| { | |
| "dataPath": "params_shard_6.bin", | |
| "format": "raw-shard", | |
| "nbytes": 45088768, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 22016 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 45088768, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c68ba999a632e735721f525e36cb4188" | |
| }, | |
| { | |
| "dataPath": "params_shard_7.bin", | |
| "format": "raw-shard", | |
| "nbytes": 25165824, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 12288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 25165824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "567649c94813883b48ed1dfcc8ee6998" | |
| }, | |
| { | |
| "dataPath": "params_shard_8.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32063488, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1048576, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 22016 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 5636096, | |
| "byteOffset": 1048576 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 1376, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 22544384, | |
| "byteOffset": 6684672 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 344, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2818048, | |
| "byteOffset": 29229056 | |
| }, | |
| { | |
| "name": "model.layers.1.input_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32047104 | |
| }, | |
| { | |
| "name": "model.layers.1.post_attention_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32055296 | |
| } | |
| ], | |
| "md5sum": "573dff167a868366a699177dbfbc44c2" | |
| }, | |
| { | |
| "dataPath": "params_shard_9.bin", | |
| "format": "raw-shard", | |
| "nbytes": 45088768, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 22016 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 45088768, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "d369ce6ca2712ad0265c77bc31aca5c9" | |
| }, | |
| { | |
| "dataPath": "params_shard_10.bin", | |
| "format": "raw-shard", | |
| "nbytes": 22544384, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 1376, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 22544384, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c642cf007bf1c783ce4318c03dd865f5" | |
| }, | |
| { | |
| "dataPath": "params_shard_11.bin", | |
| "format": "raw-shard", | |
| "nbytes": 25165824, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 12288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 25165824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c677dbc5d6763d852ccc339f40d70a46" | |
| }, | |
| { | |
| "dataPath": "params_shard_12.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32587776, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 12288 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3145728, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8388608, | |
| "byteOffset": 3145728 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1048576, | |
| "byteOffset": 11534336 | |
| }, | |
| { | |
| "name": "model.layers.2.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 22016 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 5636096, | |
| "byteOffset": 12582912 | |
| }, | |
| { | |
| "name": "model.layers.2.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 344, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2818048, | |
| "byteOffset": 18219008 | |
| }, | |
| { | |
| "name": "model.layers.2.input_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 21037056 | |
| }, | |
| { | |
| "name": "model.layers.2.post_attention_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 21045248 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 12288 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3145728, | |
| "byteOffset": 21053440 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8388608, | |
| "byteOffset": 24199168 | |
| } | |
| ], | |
| "md5sum": "2fe4e2d31ded29ecbc97713c15cf3065" | |
| }, | |
| { | |
| "dataPath": "params_shard_13.bin", | |
| "format": "raw-shard", | |
| "nbytes": 45088768, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 22016 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 45088768, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "b245482ff81bc17b05d1deaba369fddc" | |
| }, | |
| { | |
| "dataPath": "params_shard_14.bin", | |
| "format": "raw-shard", | |
| "nbytes": 25165824, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 12288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 25165824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3c69628ac1609e53340a59ba0d171bf9" | |
| }, | |
| { | |
| "dataPath": "params_shard_15.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32063488, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1048576, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 22016 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 5636096, | |
| "byteOffset": 1048576 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 1376, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 22544384, | |
| "byteOffset": 6684672 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 344, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2818048, | |
| "byteOffset": 29229056 | |
| }, | |
| { | |
| "name": "model.layers.3.input_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32047104 | |
| }, | |
| { | |
| "name": "model.layers.3.post_attention_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32055296 | |
| } | |
| ], | |
| "md5sum": "756331ddf99ea961d49b8df7b6008ed1" | |
| }, | |
| { | |
| "dataPath": "params_shard_16.bin", | |
| "format": "raw-shard", | |
| "nbytes": 45088768, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 22016 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 45088768, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "faaae6a9aea2564ef9c18b82b6cc119c" | |
| }, | |
| { | |
| "dataPath": "params_shard_17.bin", | |
| "format": "raw-shard", | |
| "nbytes": 22544384, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 1376, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 22544384, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "325c866363e7b74b4ab16daf82c0c166" | |
| }, | |
| { | |
| "dataPath": "params_shard_18.bin", | |
| "format": "raw-shard", | |
| "nbytes": 25165824, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 12288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 25165824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "be9918024a9f24380a415b53edf2dd9f" | |
| }, | |
| { | |
| "dataPath": "params_shard_19.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32587776, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 12288 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3145728, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8388608, | |
| "byteOffset": 3145728 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1048576, | |
| "byteOffset": 11534336 | |
| }, | |
| { | |
| "name": "model.layers.4.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 22016 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 5636096, | |
| "byteOffset": 12582912 | |
| }, | |
| { | |
| "name": "model.layers.4.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 344, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2818048, | |
| "byteOffset": 18219008 | |
| }, | |
| { | |
| "name": "model.layers.4.input_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 21037056 | |
| }, | |
| { | |
| "name": "model.layers.4.post_attention_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 21045248 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 12288 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 3145728, | |
| "byteOffset": 21053440 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8388608, | |
| "byteOffset": 24199168 | |
| } | |
| ], | |
| "md5sum": "44c124c1dc4bc71dab03063e8902598f" | |
| }, | |
| { | |
| "dataPath": "params_shard_20.bin", | |
| "format": "raw-shard", | |
| "nbytes": 45088768, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 512, | |
| 22016 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 45088768, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2704865ad9cc478fcb1d00bea211f60b" | |
| }, | |
| { | |
| "dataPath": "params_shard_21.bin", | |
| "format": "raw-shard", | |
| "nbytes": 65536000, | |
| "records": [ | |
| { | |
| "name": "lm_head.q_weight", | |
| "shape": [ | |
| 512, | |
| 32000 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 65536000, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3ab5243697098da96f215d9c7fe2a7ca" | |
| }, | |
| { | |
| "dataPath": "params_shard_22.bin", | |
| "format": "raw-shard", | |
| "nbytes": 32071680, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1048576, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 128, | |
| 22016 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 5636096, | |
| "byteOffset": 1048576 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 1376, | |
| 4096 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 22544384, | |
| "byteOffset": 6684672 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 344, | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2818048, | |
| "byteOffset": 29229056 | |
| }, | |
| { | |
| "name": "model.layers.5.input_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32047104 | |
| }, | |
| { | |
| "name": "model.layers.5.post_attention_layernorm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32055296 | |
| }, | |
| { | |
| "name": "model.norm.weight", | |
| "shape": [ | |
| 4096 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192, | |
| "byteOffset": 32063488 | |
| } | |
| ], | |
| "md5sum": "ca0268efd419df23435db72a5aa6133b" | |
| }, | |
| { | |
| "dataPath": "params_shard_23.bin", | |
| "format": "raw-shard", | |
| "nbytes": 8192000, | |
| "records": [ | |
| { | |
| "name": "lm_head.q_scale", | |
| "shape": [ | |
| 128, | |
| 32000 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 8192000, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4c61f062381dc9b41123ce288cc6a887" | |
| } | |
| ] | |
| } |