| { | |
| "metadata": { | |
| "ParamSize": 243, | |
| "ParamBytes": 6015407616.0, | |
| "BitsPerParam": 16.0 | |
| }, | |
| "records": [ | |
| { | |
| "dataPath": "params_shard_0.bin", | |
| "format": "raw-shard", | |
| "nbytes": 565645824, | |
| "records": [ | |
| { | |
| "name": "lm_head.weight", | |
| "shape": [ | |
| 122753, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 565645824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "aef75d1d0652f7846413b53784ca7040" | |
| }, | |
| { | |
| "dataPath": "params_shard_1.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.36.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3a6aa879d8e652ce8d0c3d688dd4ff97" | |
| }, | |
| { | |
| "dataPath": "params_shard_2.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26551296, | |
| "records": [ | |
| { | |
| "name": "model.layers.36.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.36.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 4608 | |
| }, | |
| { | |
| "name": "model.layers.36.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 26546688 | |
| } | |
| ], | |
| "md5sum": "eea7bcb1f699b3294ab5a8237e973cb7" | |
| }, | |
| { | |
| "dataPath": "params_shard_3.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.37.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4043395be4d373c0d60d8c95d1b48f5d" | |
| }, | |
| { | |
| "dataPath": "params_shard_4.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.37.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "81b7096e5aff67c6e9004b2d86f67d11" | |
| }, | |
| { | |
| "dataPath": "params_shard_5.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.37.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "08cd3e82a205262e233a126dc1e4a583" | |
| }, | |
| { | |
| "dataPath": "params_shard_6.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.38.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "d0e3227e68f517b0d0cc7b6ac030f4f9" | |
| }, | |
| { | |
| "dataPath": "params_shard_7.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.38.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2b11dbff0c504781d6562838dc101019" | |
| }, | |
| { | |
| "dataPath": "params_shard_8.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.38.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "f9e573d5d285c87ff53bf078a8ec1402" | |
| }, | |
| { | |
| "dataPath": "params_shard_9.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.39.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "19e8fe7fb20611a70c9a0e3dfbb1e758" | |
| }, | |
| { | |
| "dataPath": "params_shard_10.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.39.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "82e5c7e86b9a98a564ef3f1c30c151fc" | |
| }, | |
| { | |
| "dataPath": "params_shard_11.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.39.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ed6f93dbbd7a173f11530330a6a6cfb7" | |
| }, | |
| { | |
| "dataPath": "params_shard_12.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.36.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.37.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.37.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.37.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.38.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.38.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.38.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.39.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.39.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "5518ed9a9b0163c68f8a178f1ae536ba" | |
| }, | |
| { | |
| "dataPath": "params_shard_13.bin", | |
| "format": "raw-shard", | |
| "nbytes": 565645824, | |
| "records": [ | |
| { | |
| "name": "model.embed_tokens.weight", | |
| "shape": [ | |
| 122753, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 565645824, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ad69be1619c5d9b3e1039625be7f8c3c" | |
| }, | |
| { | |
| "dataPath": "params_shard_14.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c8eaf425bb7563689224ad660dd96c2f" | |
| }, | |
| { | |
| "dataPath": "params_shard_15.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "26c4b4799ac64291646c03de2966abf4" | |
| }, | |
| { | |
| "dataPath": "params_shard_16.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3e42be7676ba7b2d5196194a25c8ce05" | |
| }, | |
| { | |
| "dataPath": "params_shard_17.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "47032fbaaae29854164e3067b0a8d581" | |
| }, | |
| { | |
| "dataPath": "params_shard_18.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "aafa9da37c7f40ddc19cbe1a990ae2d3" | |
| }, | |
| { | |
| "dataPath": "params_shard_19.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "5f0fc15d93c1a7fa5dc1ceefecd4e133" | |
| }, | |
| { | |
| "dataPath": "params_shard_20.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4f8904000fe8e1f21be0761b5ddadba8" | |
| }, | |
| { | |
| "dataPath": "params_shard_21.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e14c0a776241b0851bc7285061a7e146" | |
| }, | |
| { | |
| "dataPath": "params_shard_22.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3c9ddaf41012579a0f3088c31d264fc3" | |
| }, | |
| { | |
| "dataPath": "params_shard_23.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31882752, | |
| "records": [ | |
| { | |
| "name": "model.layers.39.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.norm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.0.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.0.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.1.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.1.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21256704 | |
| }, | |
| { | |
| "name": "model.layers.10.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| }, | |
| { | |
| "name": "model.layers.10.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31878144 | |
| } | |
| ], | |
| "md5sum": "67b61646c680c3e5714f9225a82caf96" | |
| }, | |
| { | |
| "dataPath": "params_shard_24.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.11.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "31cd5ea2620bb047a1dd19fb00ced7fc" | |
| }, | |
| { | |
| "dataPath": "params_shard_25.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.11.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3d4663155e61cbbbec8f3637584ca5a9" | |
| }, | |
| { | |
| "dataPath": "params_shard_26.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.11.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2f7729195eeef70f05be359e94346e96" | |
| }, | |
| { | |
| "dataPath": "params_shard_27.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.12.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "a7a3f130dde33a429b181d2d15fbcbf0" | |
| }, | |
| { | |
| "dataPath": "params_shard_28.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.12.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "0d3cff21b09a9b75bbcf0e3f32f6bd76" | |
| }, | |
| { | |
| "dataPath": "params_shard_29.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.12.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e11b0152f6b48262a1cc5d8f529292f8" | |
| }, | |
| { | |
| "dataPath": "params_shard_30.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.13.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "7c565fc1eac2cd9089f495e0139fd32c" | |
| }, | |
| { | |
| "dataPath": "params_shard_31.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.13.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3a56f0ede1648d3ca958dc70adcc4453" | |
| }, | |
| { | |
| "dataPath": "params_shard_32.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.13.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "95b20856e659151cc22db3cca9f6adc1" | |
| }, | |
| { | |
| "dataPath": "params_shard_33.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.11.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.11.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.12.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.12.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.13.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.13.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "bc783960bbbdaaab21381140ce4d35b3" | |
| }, | |
| { | |
| "dataPath": "params_shard_34.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.14.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "8d93b44b02d8edf61b88a5505852151d" | |
| }, | |
| { | |
| "dataPath": "params_shard_35.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.14.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "da0ac81ab34062bcc4e4a41aa9a0075f" | |
| }, | |
| { | |
| "dataPath": "params_shard_36.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.14.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "f4c9cca8fa0f132f9b46bedfdcd29df9" | |
| }, | |
| { | |
| "dataPath": "params_shard_37.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.15.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4716fe622deb65c5b8c884c09598bd25" | |
| }, | |
| { | |
| "dataPath": "params_shard_38.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.15.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "14052dbfd5fd8ce5d0fe1c950893508a" | |
| }, | |
| { | |
| "dataPath": "params_shard_39.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.15.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ed48d06bf19f100e09d47e2242138232" | |
| }, | |
| { | |
| "dataPath": "params_shard_40.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "a78825b99be5a44c1b726cee5098299d" | |
| }, | |
| { | |
| "dataPath": "params_shard_41.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "f2edfa84b526d863dccad353b6069180" | |
| }, | |
| { | |
| "dataPath": "params_shard_42.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4a3fa9baeca996309c306e3da779f4b9" | |
| }, | |
| { | |
| "dataPath": "params_shard_43.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.13.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.14.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.14.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.15.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.15.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.16.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.16.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "4c871dfc01658f0dfe4a6e5f2987aff5" | |
| }, | |
| { | |
| "dataPath": "params_shard_44.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.17.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "97e725aeb09333c4388410f1f40f7662" | |
| }, | |
| { | |
| "dataPath": "params_shard_45.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.17.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e2f4fc481f60500feb005caaa6dea104" | |
| }, | |
| { | |
| "dataPath": "params_shard_46.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.17.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "59efccadd088c6037589dc70cfbf8fab" | |
| }, | |
| { | |
| "dataPath": "params_shard_47.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.18.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3619060e922083bf25758cbff2a3c9d3" | |
| }, | |
| { | |
| "dataPath": "params_shard_48.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.18.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "118744693a7f026d3fd54f5801d17f37" | |
| }, | |
| { | |
| "dataPath": "params_shard_49.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.18.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c8d85b2f84f3aab85617d9db3dda46e6" | |
| }, | |
| { | |
| "dataPath": "params_shard_50.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.19.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "a4e2ff932f7591885d24c42164a85bef" | |
| }, | |
| { | |
| "dataPath": "params_shard_51.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.19.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e48775efb49736234ebac4e018999bfa" | |
| }, | |
| { | |
| "dataPath": "params_shard_52.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.19.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "7bf0da7325ceebbc51fceab7b5f15b4e" | |
| }, | |
| { | |
| "dataPath": "params_shard_53.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.17.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.17.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.18.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.18.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.19.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.19.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "486f193d80686ef3872bfef2bd4b3867" | |
| }, | |
| { | |
| "dataPath": "params_shard_54.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "6c2324709548ef4ffa7284a5484512f9" | |
| }, | |
| { | |
| "dataPath": "params_shard_55.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2404e5448d21b12cdea6bb237dc337b9" | |
| }, | |
| { | |
| "dataPath": "params_shard_56.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "aed5e1efaf317f8c9e582e62db232bcc" | |
| }, | |
| { | |
| "dataPath": "params_shard_57.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.20.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "719e009dae1ccbe4384df52ea51e9233" | |
| }, | |
| { | |
| "dataPath": "params_shard_58.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.20.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "f9e647dd6ee08b73ccc87bf367c0e73d" | |
| }, | |
| { | |
| "dataPath": "params_shard_59.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.20.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "fd513ad6ad353731bc9863d99c60a6e0" | |
| }, | |
| { | |
| "dataPath": "params_shard_60.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ed42ae64165937471f1a34b96d1b1873" | |
| }, | |
| { | |
| "dataPath": "params_shard_61.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "57debfb98bc061cdce2a509f630e78f5" | |
| }, | |
| { | |
| "dataPath": "params_shard_62.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "c28e759b591746018b8104632cf1dae7" | |
| }, | |
| { | |
| "dataPath": "params_shard_63.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.19.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.2.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.2.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.20.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.20.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.21.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.21.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "1b0dbf77d69194d887c5c4f531184ec4" | |
| }, | |
| { | |
| "dataPath": "params_shard_64.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.22.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "7dbf7e1dcb3e4545ffb4010db9afb273" | |
| }, | |
| { | |
| "dataPath": "params_shard_65.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.22.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "d3158347d1cadb0a3574817058177428" | |
| }, | |
| { | |
| "dataPath": "params_shard_66.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.22.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "d6ae7a28d9c4dacbfbcfb6ed5cb83583" | |
| }, | |
| { | |
| "dataPath": "params_shard_67.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.23.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "61c7170d148ece28a0e1079c826f7a56" | |
| }, | |
| { | |
| "dataPath": "params_shard_68.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.23.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "30c5e41beb2c7680a956cb1a3dc0ba05" | |
| }, | |
| { | |
| "dataPath": "params_shard_69.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.23.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "30c22e234689b610bf910ff5e1abfad3" | |
| }, | |
| { | |
| "dataPath": "params_shard_70.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "430c29f28ca408badc5f334abd988a99" | |
| }, | |
| { | |
| "dataPath": "params_shard_71.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2e495909eb6b6388c8e060becdbbf3bb" | |
| }, | |
| { | |
| "dataPath": "params_shard_72.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "6dcc26148bacec93708451e2e642cede" | |
| }, | |
| { | |
| "dataPath": "params_shard_73.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.22.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.22.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.23.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.23.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.24.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.24.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "dd3cbd945ef771cf1254b796a4c96038" | |
| }, | |
| { | |
| "dataPath": "params_shard_74.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.25.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "1b2098818fa11b08387d1e4622281971" | |
| }, | |
| { | |
| "dataPath": "params_shard_75.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.25.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "71e47481af972c1d4147ff882e41ec90" | |
| }, | |
| { | |
| "dataPath": "params_shard_76.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.25.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e4ebf646951fbf7f9bb43d68714c71d3" | |
| }, | |
| { | |
| "dataPath": "params_shard_77.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.26.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "efe30fb9cba358d13ce26f668185303b" | |
| }, | |
| { | |
| "dataPath": "params_shard_78.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.26.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "87b1f822c80760e85c06a3264cd798ae" | |
| }, | |
| { | |
| "dataPath": "params_shard_79.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.26.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "635eb577ae1d525de08a7adc5f1013c2" | |
| }, | |
| { | |
| "dataPath": "params_shard_80.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.27.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "799c4a987d356dfd2d148b6b6aa1ceb2" | |
| }, | |
| { | |
| "dataPath": "params_shard_81.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.27.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "b5b2045368c5cbc6bbfaf842094b5064" | |
| }, | |
| { | |
| "dataPath": "params_shard_82.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.27.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "5a66f7a4f2917273d2cdf4fc30ecbccb" | |
| }, | |
| { | |
| "dataPath": "params_shard_83.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.25.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.25.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.25.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.26.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.26.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.26.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.27.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.27.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "9338f197654aa6f5e12ae6e80ced5456" | |
| }, | |
| { | |
| "dataPath": "params_shard_84.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.28.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "a3d1298b00a97148cef899e9bc5b8936" | |
| }, | |
| { | |
| "dataPath": "params_shard_85.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.28.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "7b31c1d197edbf6635efb8cb2c47c493" | |
| }, | |
| { | |
| "dataPath": "params_shard_86.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.28.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "06213f7e0e45afcb72143a5276f2b00c" | |
| }, | |
| { | |
| "dataPath": "params_shard_87.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.29.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "6af723aa213b15f5943f8e041d9b3aed" | |
| }, | |
| { | |
| "dataPath": "params_shard_88.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.29.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "acecd0b433d4a7139fba4596f7580f6f" | |
| }, | |
| { | |
| "dataPath": "params_shard_89.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.29.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "22686c74f474b0c9ba375626034eef8c" | |
| }, | |
| { | |
| "dataPath": "params_shard_90.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "75129c3f959b298db8ff6fd16f9e8b22" | |
| }, | |
| { | |
| "dataPath": "params_shard_91.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "56727363cef1a70a57084fb3cca547be" | |
| }, | |
| { | |
| "dataPath": "params_shard_92.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "7faeb7e5f251d022b643dda09dfff5cf" | |
| }, | |
| { | |
| "dataPath": "params_shard_93.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.27.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.28.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.28.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.28.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.29.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.29.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.29.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.3.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.3.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "8e96138b3df2064f1ba6a9cd66271ec1" | |
| }, | |
| { | |
| "dataPath": "params_shard_94.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.30.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "16593dcac92db8b7b1beaa44cd06238a" | |
| }, | |
| { | |
| "dataPath": "params_shard_95.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.30.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "88fe65fbe1f8506c2be17b2ec40e6a57" | |
| }, | |
| { | |
| "dataPath": "params_shard_96.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.30.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "cfb2a888aec416076077499377e98ffd" | |
| }, | |
| { | |
| "dataPath": "params_shard_97.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.31.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3b525667325049bbfddc493ed6f6936d" | |
| }, | |
| { | |
| "dataPath": "params_shard_98.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.31.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ca03e98dd65692893296902b196e1399" | |
| }, | |
| { | |
| "dataPath": "params_shard_99.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.31.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2dcea4393d7f955f4115961769b9ef32" | |
| }, | |
| { | |
| "dataPath": "params_shard_100.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.32.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e4581b478011b0941e974596e00973ac" | |
| }, | |
| { | |
| "dataPath": "params_shard_101.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.32.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "5e2c39e9a6d61ea3830aaa7a77b60203" | |
| }, | |
| { | |
| "dataPath": "params_shard_102.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.32.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3b4dc4f8a7e8bb76011e9ff09b281522" | |
| }, | |
| { | |
| "dataPath": "params_shard_103.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.30.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.30.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.30.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.31.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.31.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.31.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.32.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.32.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "15be65ea5f519cf02bac9956a015bf10" | |
| }, | |
| { | |
| "dataPath": "params_shard_104.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.33.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "1e598675cc8d5d7d48617a96f6e35ff0" | |
| }, | |
| { | |
| "dataPath": "params_shard_105.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.33.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "2204e5b4a0620d501f310b25885552ea" | |
| }, | |
| { | |
| "dataPath": "params_shard_106.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.33.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "45e5e646b1b11e290780478644f8e2f7" | |
| }, | |
| { | |
| "dataPath": "params_shard_107.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.34.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "a10773de5438185147437e0c6cec56f8" | |
| }, | |
| { | |
| "dataPath": "params_shard_108.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.34.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e685343906a04bbe1c789a3d2a2dae75" | |
| }, | |
| { | |
| "dataPath": "params_shard_109.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.34.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "68564ac602a1e2402f80bb4e13f5ca44" | |
| }, | |
| { | |
| "dataPath": "params_shard_110.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.35.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ee156e751204aa0c8a6a7b1a9580b519" | |
| }, | |
| { | |
| "dataPath": "params_shard_111.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.35.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "1757d52bafaa54d4c3faa3a94a7b2cd0" | |
| }, | |
| { | |
| "dataPath": "params_shard_112.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.35.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4ed714f71f034dbf7019b73d2f466994" | |
| }, | |
| { | |
| "dataPath": "params_shard_113.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.32.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.33.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.33.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.33.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.34.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.34.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.34.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.35.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.35.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "05b59ebce20043d928b4bd7eee42c104" | |
| }, | |
| { | |
| "dataPath": "params_shard_114.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.36.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "77d244eebcfa5fb12f7149c57ce3dcae" | |
| }, | |
| { | |
| "dataPath": "params_shard_115.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e08148adc19ec183714600352ddd713c" | |
| }, | |
| { | |
| "dataPath": "params_shard_116.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "6cf1194e1aefb265a65047cce32e228b" | |
| }, | |
| { | |
| "dataPath": "params_shard_117.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "6fafc74233cc51a0e5b35b368e904e91" | |
| }, | |
| { | |
| "dataPath": "params_shard_118.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "b505620fd3f5308d5b1f11231d20cf59" | |
| }, | |
| { | |
| "dataPath": "params_shard_119.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "d59cd7decd29fa1bdaeb089a0b65a7bc" | |
| }, | |
| { | |
| "dataPath": "params_shard_120.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "16ba153ca209205f873320cb6dbff81e" | |
| }, | |
| { | |
| "dataPath": "params_shard_121.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "945ef52ca44175f9ad10103b1dd50237" | |
| }, | |
| { | |
| "dataPath": "params_shard_122.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "66e75e100a11c29ce0bb904f306cffec" | |
| }, | |
| { | |
| "dataPath": "params_shard_123.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "79febf0852bfcbc501228df58159f393" | |
| }, | |
| { | |
| "dataPath": "params_shard_124.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.35.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.4.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.4.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.5.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.5.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.6.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.6.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "08b777a7c9f816ab8fc3ace7a522b0b2" | |
| }, | |
| { | |
| "dataPath": "params_shard_125.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.7.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "f03816d14467ad9090b7ef84cfa8739f" | |
| }, | |
| { | |
| "dataPath": "params_shard_126.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.7.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "61e2b6113f227bacf6ecdf593fd9d91c" | |
| }, | |
| { | |
| "dataPath": "params_shard_127.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.7.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "8cf9b1ce086505106372affc603cd1f2" | |
| }, | |
| { | |
| "dataPath": "params_shard_128.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.8.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "761f728fb62837f8b59669c909a7641f" | |
| }, | |
| { | |
| "dataPath": "params_shard_129.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.8.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4da6ae0348a26524ae3e5dc81a573ef0" | |
| }, | |
| { | |
| "dataPath": "params_shard_130.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.8.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "60cd90424172f73dd1ba59999295f63b" | |
| }, | |
| { | |
| "dataPath": "params_shard_131.bin", | |
| "format": "raw-shard", | |
| "nbytes": 26542080, | |
| "records": [ | |
| { | |
| "name": "model.layers.9.mlp.down_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 5760 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 26542080, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "d3e723b55f29d011257165fa6762efe6" | |
| }, | |
| { | |
| "dataPath": "params_shard_132.bin", | |
| "format": "raw-shard", | |
| "nbytes": 53084160, | |
| "records": [ | |
| { | |
| "name": "model.layers.9.mlp.gate_up_proj.weight", | |
| "shape": [ | |
| 11520, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 53084160, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "9c24f8873f15a9f3c6877ba4fad8dca7" | |
| }, | |
| { | |
| "dataPath": "params_shard_133.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31850496, | |
| "records": [ | |
| { | |
| "name": "model.layers.9.self_attn.qkv_proj.weight", | |
| "shape": [ | |
| 6912, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 31850496, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "9df73aad99e1dafbaa08c1055aee330e" | |
| }, | |
| { | |
| "dataPath": "params_shard_134.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31878144, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.7.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.7.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.7.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10626048 | |
| }, | |
| { | |
| "name": "model.layers.8.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21242880 | |
| }, | |
| { | |
| "name": "model.layers.8.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 21247488 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.9.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31868928 | |
| }, | |
| { | |
| "name": "model.layers.9.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 31873536 | |
| } | |
| ], | |
| "md5sum": "421a8ae15ac9da2a5ef2d2815ea102a0" | |
| }, | |
| { | |
| "dataPath": "params_shard_135.bin", | |
| "format": "raw-shard", | |
| "nbytes": 10616832, | |
| "records": [ | |
| { | |
| "name": "model.layers.9.self_attn.o_proj.weight", | |
| "shape": [ | |
| 2304, | |
| 2304 | |
| ], | |
| "dtype": "float16", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "579166d1d514715abdaf4566213fd59c" | |
| } | |
| ] | |
| } |