diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3705 @@ +{ + "metadata": { + "ParamSize": 315, + "ParamBytes": 1470915072.0, + "BitsPerParam": 4.501064160740709 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 294912000, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 256000, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 294912000, + "byteOffset": 0 + } + ], + "md5sum": "48fd03144d29bb2a39add46adceeda2e" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 36864000, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 256000, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 36864000, + "byteOffset": 0 + } + ], + "md5sum": "c78990eaaa4ac35c7522b188ad02a7de" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 33182208, + "records": [ + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 4608 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10621440 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11948544 + } + ], + "md5sum": "992147da627aabe0e2dd12619e2530f5" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "868e77b22c8ee361594e8127173d15d2" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.0.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.0.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.1.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.1.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "a9312c77d5cba55055b3a9ac111fb111" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "5bcd36a4054f8c7d89cb1e304b7eced1" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "14090d68970cc45ef19df49aab35cffd" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.10.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.10.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.11.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.11.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "97191482c11889ac630fbdffdfb86b86" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "927b52d22a46434ff2c7b8f7c62b90b9" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "570252955d69f06568b3641d15917e74" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.12.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.12.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.13.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.13.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "6f432a9e973212f2d895964610a59281" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "3730933b4429b6c39d723dc801757e6b" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "c3317acbf3499a86bb61603eddc64817" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.14.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.14.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.15.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.15.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "8f5ed0c662be2f6a010ff4b21860e5c1" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "15e216f3b7e6fffe6c124d8bbe711858" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "6cbe8154f893a62c883d0055f82f562f" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.16.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.16.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.17.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.17.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "90418d504c6a0e5265815777c2552aa0" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "48e1a4ee9b42b00d5e167d313c51acb7" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "a2818000756d158b94356b8ef2041a71" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.18.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.18.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.19.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.19.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "6b28914c72971e897bf7f9383342ace4" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "5a889e65a6053358323268ca9474d6a9" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "7f48e1ee75fdef122b29dbec88ff51c8" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.2.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.2.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.20.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.20.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "fc6cf0d99801ddcf3be0df182a9132a7" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "072a27db1996ebf349c933288dad85c5" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "f5577204a7ace1b7b3e416702cdb00c2" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.21.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.21.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.22.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.22.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "2277c86ab96294ba515cbeced24474fb" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "6d2ca666bd989681862b9ea875840e88" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 31864320, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.23.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.23.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10630656 + } + ], + "md5sum": "f8e7f70754daa489cddfbe7550f69bf0" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "c93d7d716ca871e790c8c171cc61658c" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2654208 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7372800 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7962624 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10321920 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10616832 + }, + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10621440 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21238272 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22565376 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25219584 + }, + { + "name": "model.layers.3.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25224192 + }, + { + "name": "model.layers.3.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25228800 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25233408 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29952000 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30541824 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32901120 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33196032 + } + ], + "md5sum": "73ddf6ce48498c5f96e4c1d7deba8c9b" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "f28f2177ff25e120e2a9c7d3115c4cbe" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "50a27688bbfdaff8ab5104076160ef86" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.4.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.4.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.5.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.5.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "d29f54fea639df03f19a5b867026553a" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "61da1f34372e1df63863a7f8a57e2d52" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "182d00e16e78d5327747c5c246c6b8cd" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.6.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.6.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.7.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.7.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "60d494f540e42ace80bfa3fcfcbc1cdc" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "4bd97e878617f859743652d8e9ace7d0" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "6d10d44ccc9c75cd1c6f99a106b585f7" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.8.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.8.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.9.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.9.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "0f51f42071697edfa60d2805476a5c38" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 288, + 18432 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "1bf6346351c43dc1e234b21c5e089d86" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 31882752, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11943936 + }, + { + "name": "model.layers.24.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11948544 + }, + { + "name": "model.layers.24.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11953152 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11957760 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 1152, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 11962368 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 288, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 22579200 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 72, + 18432 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 23906304 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26560512 + }, + { + "name": "model.layers.25.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26565120 + }, + { + "name": "model.layers.25.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26569728 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 288, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26574336 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 72, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 31292928 + } + ], + "md5sum": "f12a639aabe001181c4c9f07c31f12ff" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 2658816, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2304 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 64, + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 2359296 + }, + { + "name": "model.norm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + } + ], + "md5sum": "eb7d81336becac090a24548906c209b1" + } + ] +} \ No newline at end of file