| { | |
| "metadata": { | |
| "ParamSize": 315, | |
| "ParamBytes": 1634780160.0, | |
| "BitsPerParam": 5.002498464347751 | |
| }, | |
| "records": [ | |
| { | |
| "dataPath": "params_shard_0.bin", | |
| "format": "raw-shard", | |
| "nbytes": 294912000, | |
| "records": [ | |
| { | |
| "name": "model.embed_tokens.q_weight", | |
| "shape": [ | |
| 256000, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912000, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "88947faa8f6be77faf8c7b1c15ecd1aa" | |
| }, | |
| { | |
| "dataPath": "params_shard_1.bin", | |
| "format": "raw-shard", | |
| "nbytes": 36864000, | |
| "records": [ | |
| { | |
| "name": "model.embed_tokens.q_scale", | |
| "shape": [ | |
| 256000, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 36864000, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "625a34c9eab976edd9064fcd0bcd959b" | |
| }, | |
| { | |
| "dataPath": "params_shard_2.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33182208, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 4608 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.0.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11948544 | |
| } | |
| ], | |
| "md5sum": "18865507fc278dcc17974d4fe00a7ce5" | |
| }, | |
| { | |
| "dataPath": "params_shard_3.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.1.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "db4e058f4a1be46389caf46403232c17" | |
| }, | |
| { | |
| "dataPath": "params_shard_4.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.0.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.0.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.0.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.0.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.0.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.1.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.1.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.1.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.1.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.1.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.1.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.10.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "0008c39abd872c467b2680485cb11a75" | |
| }, | |
| { | |
| "dataPath": "params_shard_5.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.10.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.10.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "da15c539ff10c2a3c259054c52c57814" | |
| }, | |
| { | |
| "dataPath": "params_shard_6.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.11.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "e8e54bb230de9b41f99b98d16e96a4c8" | |
| }, | |
| { | |
| "dataPath": "params_shard_7.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.10.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.10.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.10.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.10.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.10.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.10.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.10.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.10.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.11.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.11.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.11.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.11.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.11.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.11.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.11.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.11.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.12.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "20b0a1178a82afffabb9ea2ef32c3ece" | |
| }, | |
| { | |
| "dataPath": "params_shard_8.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.12.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.12.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.12.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "745ed1e5d42608bfb82a1c574a1c1165" | |
| }, | |
| { | |
| "dataPath": "params_shard_9.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.13.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "1bfbaf112e662554d01fe898f929d5dc" | |
| }, | |
| { | |
| "dataPath": "params_shard_10.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.12.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.12.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.12.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.12.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.12.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.13.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.13.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.13.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.13.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.13.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.13.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.13.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.13.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.13.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.13.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.13.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.14.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "5107daea21226888ea6c21148cca6afc" | |
| }, | |
| { | |
| "dataPath": "params_shard_11.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.14.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.14.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.14.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "68e46491d49d61a6d9c293a22a0b1486" | |
| }, | |
| { | |
| "dataPath": "params_shard_12.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.15.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "7d5acc331760e30b38e5debf18a9554b" | |
| }, | |
| { | |
| "dataPath": "params_shard_13.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.14.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.14.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.14.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.14.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.14.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.15.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.15.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.15.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.15.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.15.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.15.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.15.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.15.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.16.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "93c65d473603695de9f92d347bb165a1" | |
| }, | |
| { | |
| "dataPath": "params_shard_14.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.16.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.16.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "65eb976ce17f88882e135a5977f84706" | |
| }, | |
| { | |
| "dataPath": "params_shard_15.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.17.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "61f4277b2f3efa0b7d8863dc4f96ab2b" | |
| }, | |
| { | |
| "dataPath": "params_shard_16.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.16.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.16.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.16.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.16.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.16.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.16.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.16.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.16.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.17.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.17.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.17.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.17.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.17.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.17.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.17.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.17.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.18.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "596ff29f3bd93a02db20b9bc8ac0a211" | |
| }, | |
| { | |
| "dataPath": "params_shard_17.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.18.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.18.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.18.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "6fe9e69a494e3e305eefcefb2ce6ede7" | |
| }, | |
| { | |
| "dataPath": "params_shard_18.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.19.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "95acded993c0e0e1960c4a330bb6b7ea" | |
| }, | |
| { | |
| "dataPath": "params_shard_19.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.18.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.18.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.18.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.18.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.18.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.19.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.19.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.19.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.19.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.19.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.19.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.19.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.19.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.19.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.19.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.19.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.2.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "efc0644c0348c68666998954ea802c12" | |
| }, | |
| { | |
| "dataPath": "params_shard_20.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.2.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.2.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "81bdf047b1bf7850de0960b0e6adba76" | |
| }, | |
| { | |
| "dataPath": "params_shard_21.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.20.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "17a1e03d6067a6d7e0eae4f76a5ea00b" | |
| }, | |
| { | |
| "dataPath": "params_shard_22.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.2.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.2.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.2.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.2.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.2.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.20.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.20.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.20.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.20.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.20.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.20.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.20.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.20.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.21.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "48ca861cecb730a1f4631f5b6d4e135d" | |
| }, | |
| { | |
| "dataPath": "params_shard_23.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.21.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.21.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "67a350854c768ba3881f08da6d3da12d" | |
| }, | |
| { | |
| "dataPath": "params_shard_24.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.22.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "72eb54b508552f3c8f639101d25d1515" | |
| }, | |
| { | |
| "dataPath": "params_shard_25.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.21.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.21.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.21.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.21.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.21.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.21.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.21.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.21.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.22.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.22.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.22.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.22.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.22.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.22.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.22.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.22.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.23.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "303852a15be200f6dbedbe1fd3478dc7" | |
| }, | |
| { | |
| "dataPath": "params_shard_26.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.23.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.23.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.23.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "343313d43213121da08519bc4b89960c" | |
| }, | |
| { | |
| "dataPath": "params_shard_27.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31864320, | |
| "records": [ | |
| { | |
| "name": "model.layers.23.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.23.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.23.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.23.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.23.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.24.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 10630656 | |
| } | |
| ], | |
| "md5sum": "9e14696fff3f86677548d67d2f0a2025" | |
| }, | |
| { | |
| "dataPath": "params_shard_28.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.3.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "3d7c383b720e96f39a20a43ec490ea87" | |
| }, | |
| { | |
| "dataPath": "params_shard_29.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33200640, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.24.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.24.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7372800 | |
| }, | |
| { | |
| "name": "model.layers.24.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7962624 | |
| }, | |
| { | |
| "name": "model.layers.24.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10321920 | |
| }, | |
| { | |
| "name": "model.layers.3.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10621440 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21238272 | |
| }, | |
| { | |
| "name": "model.layers.3.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22565376 | |
| }, | |
| { | |
| "name": "model.layers.3.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25219584 | |
| }, | |
| { | |
| "name": "model.layers.3.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25224192 | |
| }, | |
| { | |
| "name": "model.layers.3.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25228800 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29952000 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30541824 | |
| }, | |
| { | |
| "name": "model.layers.3.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32901120 | |
| }, | |
| { | |
| "name": "model.layers.4.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33196032 | |
| } | |
| ], | |
| "md5sum": "e94b5ddee812c6c9a88bdba56a356f3e" | |
| }, | |
| { | |
| "dataPath": "params_shard_30.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.4.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.4.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "181fb0ff8b009c327e12678ad7988169" | |
| }, | |
| { | |
| "dataPath": "params_shard_31.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.5.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "af692012ce7c8a1b913206cdb4db7431" | |
| }, | |
| { | |
| "dataPath": "params_shard_32.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.4.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.4.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.4.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.4.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.4.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.5.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.5.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.5.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.5.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.5.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.5.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.6.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "43e60881e58fc622c96086afd458fd9b" | |
| }, | |
| { | |
| "dataPath": "params_shard_33.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.6.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.6.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "c9702f3b0c884ff51229f7451fd65b22" | |
| }, | |
| { | |
| "dataPath": "params_shard_34.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.7.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "013c581f91a69990eb621f80be8ccf1c" | |
| }, | |
| { | |
| "dataPath": "params_shard_35.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.6.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.6.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.6.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.6.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.6.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.6.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.6.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.6.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.7.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.7.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.7.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.7.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.7.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.7.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.7.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.7.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.7.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.7.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.7.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.8.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "92cdda2bcc9e4e5f8baa8621cbac7b01" | |
| }, | |
| { | |
| "dataPath": "params_shard_36.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33177600, | |
| "records": [ | |
| { | |
| "name": "model.layers.8.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.8.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.8.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 11943936 | |
| } | |
| ], | |
| "md5sum": "6283ae91aec3f64417add6d6834da8be" | |
| }, | |
| { | |
| "dataPath": "params_shard_37.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.9.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "ec91412677d38e6fd0688fc207fb6e81" | |
| }, | |
| { | |
| "dataPath": "params_shard_38.bin", | |
| "format": "raw-shard", | |
| "nbytes": 33214464, | |
| "records": [ | |
| { | |
| "name": "model.layers.8.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.8.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| }, | |
| { | |
| "name": "model.layers.8.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2658816 | |
| }, | |
| { | |
| "name": "model.layers.8.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2663424 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 2668032 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 7386624 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 7976448 | |
| }, | |
| { | |
| "name": "model.layers.8.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 10335744 | |
| }, | |
| { | |
| "name": "model.layers.9.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 10630656 | |
| }, | |
| { | |
| "name": "model.layers.9.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 10635264 | |
| }, | |
| { | |
| "name": "model.layers.9.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 21252096 | |
| }, | |
| { | |
| "name": "model.layers.9.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.9.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25233408 | |
| }, | |
| { | |
| "name": "model.layers.9.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25238016 | |
| }, | |
| { | |
| "name": "model.layers.9.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 25242624 | |
| }, | |
| { | |
| "name": "model.layers.9.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 25247232 | |
| }, | |
| { | |
| "name": "model.layers.9.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 29965824 | |
| }, | |
| { | |
| "name": "model.layers.9.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 30555648 | |
| }, | |
| { | |
| "name": "model.layers.9.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 32914944 | |
| }, | |
| { | |
| "name": "model.layers.24.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 33209856 | |
| } | |
| ], | |
| "md5sum": "ce38991c2a0a23e72d22011e2c472b18" | |
| }, | |
| { | |
| "dataPath": "params_shard_39.bin", | |
| "format": "raw-shard", | |
| "nbytes": 21233664, | |
| "records": [ | |
| { | |
| "name": "model.layers.25.mlp.gate_up_proj.q_weight", | |
| "shape": [ | |
| 18432, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 21233664, | |
| "byteOffset": 0 | |
| } | |
| ], | |
| "md5sum": "4188b693cd447d20a7e0885a2a655178" | |
| }, | |
| { | |
| "dataPath": "params_shard_40.bin", | |
| "format": "raw-shard", | |
| "nbytes": 31882752, | |
| "records": [ | |
| { | |
| "name": "model.layers.24.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.24.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 10616832 | |
| }, | |
| { | |
| "name": "model.layers.24.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 11943936 | |
| }, | |
| { | |
| "name": "model.layers.24.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 11948544 | |
| }, | |
| { | |
| "name": "model.layers.24.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 11953152 | |
| }, | |
| { | |
| "name": "model.layers.25.input_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 11957760 | |
| }, | |
| { | |
| "name": "model.layers.25.mlp.down_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 1152 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 10616832, | |
| "byteOffset": 11962368 | |
| }, | |
| { | |
| "name": "model.layers.25.mlp.down_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 288 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 1327104, | |
| "byteOffset": 22579200 | |
| }, | |
| { | |
| "name": "model.layers.25.mlp.gate_up_proj.q_scale", | |
| "shape": [ | |
| 18432, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2654208, | |
| "byteOffset": 23906304 | |
| }, | |
| { | |
| "name": "model.layers.25.post_attention_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 26560512 | |
| }, | |
| { | |
| "name": "model.layers.25.post_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 26565120 | |
| }, | |
| { | |
| "name": "model.layers.25.pre_feedforward_layernorm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 26569728 | |
| }, | |
| { | |
| "name": "model.layers.25.self_attn.qkv_proj.q_weight", | |
| "shape": [ | |
| 4096, | |
| 288 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4718592, | |
| "byteOffset": 26574336 | |
| }, | |
| { | |
| "name": "model.layers.25.self_attn.qkv_proj.q_scale", | |
| "shape": [ | |
| 4096, | |
| 72 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 589824, | |
| "byteOffset": 31292928 | |
| } | |
| ], | |
| "md5sum": "34eaf243f1fee5545431870c346cd5b6" | |
| }, | |
| { | |
| "dataPath": "params_shard_41.bin", | |
| "format": "raw-shard", | |
| "nbytes": 2658816, | |
| "records": [ | |
| { | |
| "name": "model.layers.25.self_attn.o_proj.q_weight", | |
| "shape": [ | |
| 2304, | |
| 256 | |
| ], | |
| "dtype": "uint32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 2359296, | |
| "byteOffset": 0 | |
| }, | |
| { | |
| "name": "model.layers.25.self_attn.o_proj.q_scale", | |
| "shape": [ | |
| 2304, | |
| 64 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 294912, | |
| "byteOffset": 2359296 | |
| }, | |
| { | |
| "name": "model.norm.weight", | |
| "shape": [ | |
| 2304 | |
| ], | |
| "dtype": "float32", | |
| "format": "f32-to-bf16", | |
| "nbytes": 4608, | |
| "byteOffset": 2654208 | |
| } | |
| ], | |
| "md5sum": "03761a612d582fcbd96edf400c2f1ae5" | |
| } | |
| ] | |
| } |