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": "4420a6befc764a15bc4099bb492f452a" + }, + { + "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 4608 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10621440 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11948544 + } + ], + "md5sum": "91fed24f6223d8f1c5cd262d851c67e6" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "9dc6dc6172cd29e640a17fbf285ea060" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "24c3d4bf7f6ef31e6318e1ceb598a1fb" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "9416398ef440c4992d749b132a463aff" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "1146ddd2eb0fa2568e299bcb6f891caf" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "d323bd682e4a13701ecabf5e97243583" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "ca2d73aafe72b3c43853af66aa811593" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "bf3f07064774d4a203799cd6bf4d529c" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "3a19a0bb716674f084e5f82bcbab9959" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "8fcdb25f89cd18458c65c3ac14455155" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "e99ff98637007c3564f95f9e4c79ef5e" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "c819cf87af770b208d15f58c715f49bb" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "2edc9436e7ab928720ef68a15756ae77" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "2fd4e7f765f6406f93d9cf7697112cd8" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "400d83a49a378220ddcc007832f7771c" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "ca7756d0da7f0020ed0db54b98927ddc" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "3cd62dd964eefd12cedcd3c72426eaa1" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "d575fcdb03cbae1fe2cdd0e7b1159aa5" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "283df445b1522fde59870a1d4a703563" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "6d8023176f2c78f7389bccd8c7acc87c" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "d0bcdaa9a34a680ce8f70c230727f72d" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "e9c37f2c95dc0577d9072945e65c6f6c" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "266cb32eaedb5b9e53c2823dc5d12988" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "54860af943966efc6b355c183f7115fb" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "2808fad383f2ad9155ba9a42300af1e1" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 31864320, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10630656 + } + ], + "md5sum": "308c0c601411cdcbf42b210f30ba9cbd" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "29c7409a515ce3c67ec40d62e530421d" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2654208 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7372800 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7962624 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10621440 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21238272 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25233408 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29952000 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30541824 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "f69e2445cdd670215aef1d55f91a1c38" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "9ac699261a5450e6d13800637cfa429e" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "3bcff6f5008d0f3c3972cae0066d5e26" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "5da3c2f667640b82d1fd6c52a541e996" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "c37d6b77f5f0d7305be5cd1237d96d0c" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "1f7f0a6d927d4d4c9b7be36019b9b75b" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "ca28dd5ecfdf8332a5088ab94ce7c163" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "aaa86bd5f8ec1e0d3d1e7e5ada6a0f1a" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "e3dde2981bd66da78f267adf60a96cc3" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "1770aa95d267c4d92ff446e04b4322f2" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "ccf961fb06e686a9a20ee5b2441f9a2f" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 31882752, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "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": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 11962368 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 22579200 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "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": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26574336 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 31292928 + } + ], + "md5sum": "1abd9defd452cfff3bf065283fe55283" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 2658816, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "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": "419b019e49ae17d5556bdfdd59bab79d" + } + ] +} \ No newline at end of file