|
{ |
|
"metadata": { |
|
"ParamSize": 445, |
|
"ParamBytes": 2222822400.0, |
|
"BitsPerParam": 4.501497946035061 |
|
}, |
|
"records": [ |
|
{ |
|
"dataPath": "params_shard_0.bin", |
|
"format": "raw-shard", |
|
"nbytes": 194478080, |
|
"records": [ |
|
{ |
|
"name": "lm_head.q_weight", |
|
"shape": [ |
|
320, |
|
151936 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 194478080, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "3190c191fd717462fa3bd1089c328b50" |
|
}, |
|
{ |
|
"dataPath": "params_shard_1.bin", |
|
"format": "raw-shard", |
|
"nbytes": 33162240, |
|
"records": [ |
|
{ |
|
"name": "lm_head.q_scale", |
|
"shape": [ |
|
80, |
|
151936 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 24309760, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.20.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 24309760 |
|
}, |
|
{ |
|
"name": "model.layers.20.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 24314880 |
|
} |
|
], |
|
"md5sum": "dc70a2fc78de2e7fabf1638f149d489c" |
|
}, |
|
{ |
|
"dataPath": "params_shard_2.bin", |
|
"format": "raw-shard", |
|
"nbytes": 24709120, |
|
"records": [ |
|
{ |
|
"name": "model.layers.20.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.20.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 1105920 |
|
}, |
|
{ |
|
"name": "model.layers.20.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 18800640 |
|
}, |
|
{ |
|
"name": "model.layers.20.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 21012480 |
|
}, |
|
{ |
|
"name": "model.layers.20.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 21017600 |
|
}, |
|
{ |
|
"name": "model.layers.20.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 24294400 |
|
}, |
|
{ |
|
"name": "model.layers.21.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 24704000 |
|
} |
|
], |
|
"md5sum": "8e059664722fc6cb624e6aa7c68e7158" |
|
}, |
|
{ |
|
"dataPath": "params_shard_3.bin", |
|
"format": "raw-shard", |
|
"nbytes": 29880320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.21.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.21.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 8847360 |
|
}, |
|
{ |
|
"name": "model.layers.21.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 9953280 |
|
}, |
|
{ |
|
"name": "model.layers.21.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 27648000 |
|
}, |
|
{ |
|
"name": "model.layers.21.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 29859840 |
|
}, |
|
{ |
|
"name": "model.layers.21.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 29864960 |
|
} |
|
], |
|
"md5sum": "adee6594f063f27b48019bc42f9bf232" |
|
}, |
|
{ |
|
"dataPath": "params_shard_4.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.22.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "4f7ce23b911913d3210223affb6e140d" |
|
}, |
|
{ |
|
"dataPath": "params_shard_5.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.21.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.21.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.21.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.21.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.22.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.22.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.22.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.22.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.22.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.22.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "23aa3d25cfb8e9cb457f151bdb8d2e0e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_6.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.23.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "a3c1dc327cfd5be3d8c7af17687a9bfc" |
|
}, |
|
{ |
|
"dataPath": "params_shard_7.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.22.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.22.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.22.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.22.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.23.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.23.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.23.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.23.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.23.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.23.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "926f2b08da73a90bdf125ee5bee11d84" |
|
}, |
|
{ |
|
"dataPath": "params_shard_8.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.24.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "50c6eecc059400b818665ec20b1b75fe" |
|
}, |
|
{ |
|
"dataPath": "params_shard_9.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.23.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.23.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.23.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.23.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.24.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.24.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.24.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.24.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.24.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.24.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "6c80dd8b48326ce40a957a9536f70392" |
|
}, |
|
{ |
|
"dataPath": "params_shard_10.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.25.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "688b6b767a2205d7672b77780f5d5a45" |
|
}, |
|
{ |
|
"dataPath": "params_shard_11.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.24.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.24.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.24.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.24.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.25.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.25.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.25.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.25.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.25.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.25.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "ad46e470df7d16fbaa5fb24fc05210d3" |
|
}, |
|
{ |
|
"dataPath": "params_shard_12.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.26.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "5643299e1d44ddbc4cc82067714c722f" |
|
}, |
|
{ |
|
"dataPath": "params_shard_13.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.25.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.25.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.25.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.25.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.26.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.26.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.26.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.26.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.26.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.26.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "53766779497aeec868627b8b9913f50e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_14.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.27.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "f9e61301bfb77763eb25d99ea7f95432" |
|
}, |
|
{ |
|
"dataPath": "params_shard_15.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.26.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.26.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.26.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.26.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.27.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.27.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.27.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.27.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.27.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.27.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "8c6eac1b0e58ac046a5ad249ffa00f65" |
|
}, |
|
{ |
|
"dataPath": "params_shard_16.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.28.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "cffa2a9cc3dc73cbbd2e556f439d1fcc" |
|
}, |
|
{ |
|
"dataPath": "params_shard_17.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.27.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.27.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.27.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.27.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.28.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.28.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.28.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.28.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.28.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.28.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "73d9170fe16f91a44a2458adee2d15d2" |
|
}, |
|
{ |
|
"dataPath": "params_shard_18.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.29.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "428dc98f9a276955256d28fe2f77a3fe" |
|
}, |
|
{ |
|
"dataPath": "params_shard_19.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.28.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.28.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.28.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.28.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.29.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.29.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.29.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.29.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.29.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.29.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "20427eb6f665abea4ccf279295c06f1f" |
|
}, |
|
{ |
|
"dataPath": "params_shard_20.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.30.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "5d85aeece5673a693641f2a9aab796b4" |
|
}, |
|
{ |
|
"dataPath": "params_shard_21.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.29.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.29.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.29.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.29.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.30.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.30.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.30.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.30.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.30.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.30.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "091c2b6173a8ecdcec1d44fa877c236e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_22.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.31.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "a84a540b98018e0442fef29039f65eaf" |
|
}, |
|
{ |
|
"dataPath": "params_shard_23.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.30.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.30.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.30.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.30.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.31.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.31.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.31.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.31.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.31.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.31.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "dc07107fa0b1951be75ff43fece6423e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_24.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.32.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "c1f6d600dd914c17b4bb73b83635ee01" |
|
}, |
|
{ |
|
"dataPath": "params_shard_25.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.31.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.31.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.31.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.31.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.32.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.32.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.32.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.32.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.32.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.32.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "5bc7b9f1861bad267c9c9a769ce3d177" |
|
}, |
|
{ |
|
"dataPath": "params_shard_26.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.33.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "0ec994665b3ff0f21cbe35cfb93a9c55" |
|
}, |
|
{ |
|
"dataPath": "params_shard_27.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.32.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.32.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.32.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.32.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.33.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.33.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.33.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.33.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.33.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.33.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "b6cd6955495b3e6992b03bccbe8b6291" |
|
}, |
|
{ |
|
"dataPath": "params_shard_28.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.34.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "7858c1d5f41f1aed2b982c77f30cf7be" |
|
}, |
|
{ |
|
"dataPath": "params_shard_29.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.33.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.33.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.33.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.33.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.34.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.34.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.34.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.34.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.34.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.34.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "b19ad16cadbdce9f8aa6e90bca777f28" |
|
}, |
|
{ |
|
"dataPath": "params_shard_30.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.35.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "e60c3a238f1c62929d90aab7d5702b97" |
|
}, |
|
{ |
|
"dataPath": "params_shard_31.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.34.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.34.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.34.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.34.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.35.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.35.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.35.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.35.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.35.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.35.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "1c02f14ac678e978376af610856d0c9f" |
|
}, |
|
{ |
|
"dataPath": "params_shard_32.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.36.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "3114c6ba3fd412522b9f32681267de59" |
|
}, |
|
{ |
|
"dataPath": "params_shard_33.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.35.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.35.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.35.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.35.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.36.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.36.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.36.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.36.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.36.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.36.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "07d3c778be8f095e857609c4e4250a00" |
|
}, |
|
{ |
|
"dataPath": "params_shard_34.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.37.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "fb7f93fff36da29d31b5aea46cb60d2a" |
|
}, |
|
{ |
|
"dataPath": "params_shard_35.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.36.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.36.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.36.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.36.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.37.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.37.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.37.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.37.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.37.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.37.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "7546c535786f98b03fa538c5acfb51d5" |
|
}, |
|
{ |
|
"dataPath": "params_shard_36.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.38.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "74757792d4658a9ff3b6e2c0d54ef050" |
|
}, |
|
{ |
|
"dataPath": "params_shard_37.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.37.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.37.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.37.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.37.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.38.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.38.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.38.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.38.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.38.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.38.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "73389954779ddb637b789ad1187451b3" |
|
}, |
|
{ |
|
"dataPath": "params_shard_38.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.39.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "004025ffcdc8c83293ca54a38500f262" |
|
}, |
|
{ |
|
"dataPath": "params_shard_39.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.38.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.38.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.38.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.38.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.39.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.39.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.39.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.39.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.39.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.39.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "cdab2d919050146bf8109eb1567ef2ee" |
|
}, |
|
{ |
|
"dataPath": "params_shard_40.bin", |
|
"format": "raw-shard", |
|
"nbytes": 194478080, |
|
"records": [ |
|
{ |
|
"name": "model.embed_tokens.q_weight", |
|
"shape": [ |
|
151936, |
|
320 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 194478080, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "6ebd7ba77c46250f801bccadf2df0409" |
|
}, |
|
{ |
|
"dataPath": "params_shard_41.bin", |
|
"format": "raw-shard", |
|
"nbytes": 24309760, |
|
"records": [ |
|
{ |
|
"name": "model.embed_tokens.q_scale", |
|
"shape": [ |
|
151936, |
|
80 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 24309760, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "8a845e88f96a3273ec4caec228319c2d" |
|
}, |
|
{ |
|
"dataPath": "params_shard_42.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.0.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "019348413564dc72a717cffe190e617b" |
|
}, |
|
{ |
|
"dataPath": "params_shard_43.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26941440, |
|
"records": [ |
|
{ |
|
"name": "model.layers.39.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.39.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.39.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.39.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.norm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.0.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.0.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14755840 |
|
}, |
|
{ |
|
"name": "model.layers.0.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23603200 |
|
}, |
|
{ |
|
"name": "model.layers.0.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24709120 |
|
}, |
|
{ |
|
"name": "model.layers.0.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26920960 |
|
}, |
|
{ |
|
"name": "model.layers.0.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26926080 |
|
} |
|
], |
|
"md5sum": "ece807e4e434a09ed8ebad54daea6fb5" |
|
}, |
|
{ |
|
"dataPath": "params_shard_44.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.1.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "506ce81c56876f4b19b1694e1c558db9" |
|
}, |
|
{ |
|
"dataPath": "params_shard_45.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.0.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.0.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.0.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.0.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.1.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.1.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.1.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.1.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.1.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.1.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "95e233cd74262d9793faea2bf4baf599" |
|
}, |
|
{ |
|
"dataPath": "params_shard_46.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.10.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "10ee9431f33585abfc066712bb2c8c94" |
|
}, |
|
{ |
|
"dataPath": "params_shard_47.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.1.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.1.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.1.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.1.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.10.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.10.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.10.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.10.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.10.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.10.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "31adbac35d39c2cc78efb333d38db582" |
|
}, |
|
{ |
|
"dataPath": "params_shard_48.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.11.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "b423138cdc5cda7bdb447e816ab021dd" |
|
}, |
|
{ |
|
"dataPath": "params_shard_49.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.10.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.10.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.10.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.10.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.11.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.11.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.11.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.11.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.11.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.11.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "7f885a71abc464913a86c807bfbb17b9" |
|
}, |
|
{ |
|
"dataPath": "params_shard_50.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.12.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "eeba9cc259a4bf277a499ed3ca2f6dd2" |
|
}, |
|
{ |
|
"dataPath": "params_shard_51.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.11.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.11.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.11.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.11.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.12.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.12.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.12.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.12.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.12.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.12.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "8b76db922b28425840ab59b90d10b50e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_52.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.13.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "21c417e846c795927d9240286c66b1f8" |
|
}, |
|
{ |
|
"dataPath": "params_shard_53.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.12.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.12.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.12.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.12.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.13.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.13.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.13.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.13.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.13.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.13.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "ba78dbcf89fef20f90f594d134328556" |
|
}, |
|
{ |
|
"dataPath": "params_shard_54.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.14.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "cff9d0e2a764a47403bc2e73d500db12" |
|
}, |
|
{ |
|
"dataPath": "params_shard_55.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.13.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.13.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.13.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.13.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.14.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.14.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.14.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.14.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.14.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.14.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "e1eb9ce1b31da53b6043f0f55904bace" |
|
}, |
|
{ |
|
"dataPath": "params_shard_56.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.15.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "63503437291a1aa6f9f3260adef2b430" |
|
}, |
|
{ |
|
"dataPath": "params_shard_57.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.14.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.14.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.14.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.14.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.15.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.15.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.15.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.15.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.15.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.15.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "243732889142e53c1ffd093496b89f46" |
|
}, |
|
{ |
|
"dataPath": "params_shard_58.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.16.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "a33fb11044437264ec44c4571d73683c" |
|
}, |
|
{ |
|
"dataPath": "params_shard_59.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.15.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.15.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.15.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.15.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.16.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.16.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.16.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.16.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.16.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.16.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "48558c93e204ddfe84177fe67b1599cd" |
|
}, |
|
{ |
|
"dataPath": "params_shard_60.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.17.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "61f85543d71d1263af15c62dbcd2a717" |
|
}, |
|
{ |
|
"dataPath": "params_shard_61.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.16.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.16.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.16.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.16.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.17.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.17.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.17.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.17.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.17.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.17.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "bdc4eeac7e2680d91dc9c6eab7c49580" |
|
}, |
|
{ |
|
"dataPath": "params_shard_62.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.18.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "b7352462d090014ad76a1ec85b982923" |
|
}, |
|
{ |
|
"dataPath": "params_shard_63.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.17.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.17.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.17.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.17.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.18.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.18.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.18.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.18.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.18.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.18.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "d807f79efc58cb4f5eba34969b829202" |
|
}, |
|
{ |
|
"dataPath": "params_shard_64.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.19.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "df2aa7b7052dfede0d880918c79335e9" |
|
}, |
|
{ |
|
"dataPath": "params_shard_65.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.18.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.18.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.18.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.18.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.19.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.19.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.19.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.19.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.19.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.19.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "2b9be1894525cb9cdf4694851c675f0e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_66.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.2.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "24195fbd8f56dde808fbab02657fe11b" |
|
}, |
|
{ |
|
"dataPath": "params_shard_67.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.19.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.19.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.19.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.19.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.2.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.2.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.2.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.2.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.2.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.2.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "34673f8856493aa0dbcb25b6940e231d" |
|
}, |
|
{ |
|
"dataPath": "params_shard_68.bin", |
|
"format": "raw-shard", |
|
"nbytes": 25825280, |
|
"records": [ |
|
{ |
|
"name": "model.layers.2.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.2.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.2.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.2.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.20.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.20.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 14760960 |
|
}, |
|
{ |
|
"name": "model.layers.20.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 24591360 |
|
}, |
|
{ |
|
"name": "model.layers.3.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 25820160 |
|
} |
|
], |
|
"md5sum": "3e9e487e54aa948866ff3712810fd4e0" |
|
}, |
|
{ |
|
"dataPath": "params_shard_69.bin", |
|
"format": "raw-shard", |
|
"nbytes": 29880320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.3.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.3.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 8847360 |
|
}, |
|
{ |
|
"name": "model.layers.3.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 9953280 |
|
}, |
|
{ |
|
"name": "model.layers.3.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 27648000 |
|
}, |
|
{ |
|
"name": "model.layers.3.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 29859840 |
|
}, |
|
{ |
|
"name": "model.layers.3.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 29864960 |
|
} |
|
], |
|
"md5sum": "edebff3da923812bb5f5c4dda7abc4b0" |
|
}, |
|
{ |
|
"dataPath": "params_shard_70.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.4.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "c5c7fc253038954e814916b438ebb84e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_71.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.3.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.3.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.3.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.3.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.4.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.4.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.4.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.4.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.4.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.4.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "df75706dbb8c9074784b47536d475c9f" |
|
}, |
|
{ |
|
"dataPath": "params_shard_72.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.5.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "c8fc9db0a2d344c24824d3059a4ad171" |
|
}, |
|
{ |
|
"dataPath": "params_shard_73.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.4.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.4.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.4.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.4.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.5.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.5.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.5.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.5.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.5.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.5.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "6ae94e67224b9b27a21077f7c1402c84" |
|
}, |
|
{ |
|
"dataPath": "params_shard_74.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.6.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "658ee5f4aa3a83fd01e39677db1ff132" |
|
}, |
|
{ |
|
"dataPath": "params_shard_75.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.5.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.5.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.5.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.5.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.6.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.6.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.6.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.6.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.6.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.6.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "b2170f1b09548d62461705b07bee0c9a" |
|
}, |
|
{ |
|
"dataPath": "params_shard_76.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.7.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "97cbe13579c97ced36e670338fdc640b" |
|
}, |
|
{ |
|
"dataPath": "params_shard_77.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.6.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.6.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.6.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.6.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.7.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.7.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.7.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.7.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.7.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.7.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "a49fe30c52c526d0fe6bd2b5336530b3" |
|
}, |
|
{ |
|
"dataPath": "params_shard_78.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.8.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "e736686e685a2587ebabb74296e97f62" |
|
}, |
|
{ |
|
"dataPath": "params_shard_79.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.7.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.7.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.7.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.7.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.8.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.8.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.8.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.8.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.8.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.8.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "01a779e8b88951d727d4257eaf487e6b" |
|
}, |
|
{ |
|
"dataPath": "params_shard_80.bin", |
|
"format": "raw-shard", |
|
"nbytes": 17694720, |
|
"records": [ |
|
{ |
|
"name": "model.layers.9.mlp.gate_up_proj.q_weight", |
|
"shape": [ |
|
320, |
|
13824 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 17694720, |
|
"byteOffset": 0 |
|
} |
|
], |
|
"md5sum": "c6a3e87ab5e735f9a25dc7dd0fcf9f26" |
|
}, |
|
{ |
|
"dataPath": "params_shard_81.bin", |
|
"format": "raw-shard", |
|
"nbytes": 26936320, |
|
"records": [ |
|
{ |
|
"name": "model.layers.8.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.8.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.8.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.8.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
}, |
|
{ |
|
"name": "model.layers.9.input_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 14745600 |
|
}, |
|
{ |
|
"name": "model.layers.9.mlp.down_proj.q_weight", |
|
"shape": [ |
|
864, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 8847360, |
|
"byteOffset": 14750720 |
|
}, |
|
{ |
|
"name": "model.layers.9.mlp.down_proj.q_scale", |
|
"shape": [ |
|
216, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1105920, |
|
"byteOffset": 23598080 |
|
}, |
|
{ |
|
"name": "model.layers.9.mlp.gate_up_proj.q_scale", |
|
"shape": [ |
|
80, |
|
13824 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 2211840, |
|
"byteOffset": 24704000 |
|
}, |
|
{ |
|
"name": "model.layers.9.post_attention_layernorm.weight", |
|
"shape": [ |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 5120, |
|
"byteOffset": 26915840 |
|
}, |
|
{ |
|
"name": "model.layers.9.self_attn.c_attn.bias", |
|
"shape": [ |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 15360, |
|
"byteOffset": 26920960 |
|
} |
|
], |
|
"md5sum": "b1327580a8790432954c60b4b058546e" |
|
}, |
|
{ |
|
"dataPath": "params_shard_82.bin", |
|
"format": "raw-shard", |
|
"nbytes": 14745600, |
|
"records": [ |
|
{ |
|
"name": "model.layers.9.self_attn.c_attn.q_weight", |
|
"shape": [ |
|
320, |
|
7680 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 9830400, |
|
"byteOffset": 0 |
|
}, |
|
{ |
|
"name": "model.layers.9.self_attn.c_attn.q_scale", |
|
"shape": [ |
|
80, |
|
7680 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 1228800, |
|
"byteOffset": 9830400 |
|
}, |
|
{ |
|
"name": "model.layers.9.self_attn.o_proj.q_weight", |
|
"shape": [ |
|
320, |
|
2560 |
|
], |
|
"dtype": "uint32", |
|
"format": "f32-to-bf16", |
|
"nbytes": 3276800, |
|
"byteOffset": 11059200 |
|
}, |
|
{ |
|
"name": "model.layers.9.self_attn.o_proj.q_scale", |
|
"shape": [ |
|
80, |
|
2560 |
|
], |
|
"dtype": "float16", |
|
"format": "f32-to-bf16", |
|
"nbytes": 409600, |
|
"byteOffset": 14336000 |
|
} |
|
], |
|
"md5sum": "58327bd6d79e22c95d29f337c00bc3f1" |
|
} |
|
] |
|
} |