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": 1634780160.0, + "BitsPerParam": 5.002498464347751 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 294912000, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 256000, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 294912000, + "byteOffset": 0 + } + ], + "md5sum": "612d422b371efae782b9bdf587d456c2" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 36864000, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 256000, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 36864000, + "byteOffset": 0 + } + ], + "md5sum": "05d3669cf2f095284d6a8cc72492d0be" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 33182208, + "records": [ + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 4608 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10621440 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11948544 + } + ], + "md5sum": "895c5033226f9ecd675fba8cb071c5ed" + }, + { + "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": "47b078495c7b41f3dbe93abcc9e783c3" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.0.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.0.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.1.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.1.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "e51fe212362140d229a98156296b21d7" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "2db9a677c9638d352712f830a1187a63" + }, + { + "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": "f73939fb2b086eced95df5ede781e5ea" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.10.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.10.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.11.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.11.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "12f0f5baab0a6a74ed70904f73e86834" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "30172ce168f4d9d708e154585b68d92e" + }, + { + "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": "7c890894ca00a0e0b5cbc39f60727444" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.12.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.12.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.13.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.13.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "1dac0df10cc755db26c8bb84bd64f980" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "3c3e5a974df2960ec17869073f0c5834" + }, + { + "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": "f2f014be65ec6fa95d84ea917625fc9a" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.14.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.14.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.15.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.15.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "4bc587ec19c501cac08c376aede4aede" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "88ee55fde57528a16fb6d172b34ebf3d" + }, + { + "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": "11e847dfc5cc3f9676e605e19793d35e" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.16.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.16.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.17.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.17.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "23dc035242120ef58ce201432e871afe" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "8b5790a7fbaec484143fb73ab9ff1b87" + }, + { + "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": "78cc1a8544dd3dd9f714882897015732" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.18.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.18.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.19.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.19.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "c089b8c1f58adcdb6ff17d4913e18da2" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "d672cf18b21d5e04d9e10c911e6ce202" + }, + { + "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": "11ea59d938afb42284a4a75da814f68f" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.2.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.2.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.20.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.20.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "a46f0352022ec5263620d7f050cb7f28" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "235f998475292de9be4f91bb8d8af2e4" + }, + { + "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": "0e0fbd205015f0624432f64ac0e92610" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.21.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.21.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.22.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.22.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "84086688f6883d5ca4f0a8ca31c72ba8" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "92534c70c4fb94db3dd04632676fd3ef" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 31864320, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.23.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.23.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10630656 + } + ], + "md5sum": "cafde2104389d11e4b3b746fce7e2c28" + }, + { + "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": "c1f15e00bf9f4960d36773087eb2385c" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2654208 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7372800 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7962624 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10321920 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10616832 + }, + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10621440 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21238272 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22565376 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25219584 + }, + { + "name": "model.layers.3.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25224192 + }, + { + "name": "model.layers.3.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25228800 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25233408 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29952000 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30541824 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32901120 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33196032 + } + ], + "md5sum": "a1511a7e8997634428017a015e9fe52c" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "c3bba1611728b16c60a0cf30b709e8dd" + }, + { + "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": "613fcc71ce2615204d5d9fbb10bd2e67" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.4.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.4.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.5.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.5.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "87eac0b55e8d25592a291e01fe590b6a" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "9b5cdf79ef9ffd3450bb74be01956297" + }, + { + "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": "bc3f5991e758e5e5d967275ae02b88b5" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.6.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.6.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.7.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.7.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "2cc576514f4aec38f91113971615f17a" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 33177600, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 11943936 + } + ], + "md5sum": "cb44ad8f6668874750833a56d916b3a3" + }, + { + "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": "c34cbb6ca8d93c66cbc1e1c6e5aea698" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 33214464, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 0 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + }, + { + "name": "model.layers.8.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2658816 + }, + { + "name": "model.layers.8.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2663424 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 2668032 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 7386624 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 7976448 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 10335744 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 10630656 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 10635264 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 21252096 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 22579200 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25233408 + }, + { + "name": "model.layers.9.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25238016 + }, + { + "name": "model.layers.9.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 25242624 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 25247232 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 29965824 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 30555648 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 32914944 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 33209856 + } + ], + "md5sum": "51b0b1f9c23988b17da5dabf97282999" + }, + { + "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": "cf0b41e23e186027e33dbfdf7ff39ca0" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 31882752, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 10616832 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11943936 + }, + { + "name": "model.layers.24.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11948544 + }, + { + "name": "model.layers.24.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11953152 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 11957760 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 11962368 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 2304, + 288 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1327104, + "byteOffset": 22579200 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 2654208, + "byteOffset": 23906304 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26560512 + }, + { + "name": "model.layers.25.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26565120 + }, + { + "name": "model.layers.25.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26569728 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26574336 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 72 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 31292928 + } + ], + "md5sum": "6fd8ff001722ef26ae329347ea5cac87" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 2658816, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 64 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 294912, + "byteOffset": 2359296 + }, + { + "name": "model.norm.weight", + "shape": [ + 2304 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 2654208 + } + ], + "md5sum": "ce3537a5450fc4f0ed6b15a241828e0b" + } + ] +} \ No newline at end of file