diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3697 @@ +{ + "metadata": { + "ParamSize": 315, + "ParamBytes": 1348379136.0, + "BitsPerParam": 4.076776454709711 + }, + "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": "a17ddd4ac7972efd6b46a7fee9296a06" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "02f5141225c696855f8ca2dec7d3f25f" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 28150272, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 256000, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9216000, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 9216000 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 9220608 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 19837440 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 20169216 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 20832768 + }, + { + "name": "model.layers.0.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 20837376 + }, + { + "name": "model.layers.0.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 20841984 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 20846592 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 25565184 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 25712640 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 28071936 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 28145664 + } + ], + "md5sum": "4d8977aeab5d5df0c8f0f9f8c176ea22" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.1.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.1.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "3614e45271e555113c5775208893712f" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "35f1b927c8b50a6943c4ec3a947af78e" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.10.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.10.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "06f00d1656cc103d170bfef2ab2bc458" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.11.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.11.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "977ff36a56f8c93348c64595ca159673" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "62cebd023f3a02eceb8224784a5bea8b" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.12.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.12.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "f41deaa36cdf8c4aa3312c7272a46015" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.13.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.13.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "ccbd8edbb8e517f916ac7bf500b67133" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "016c35617d7f4154c03f143790890687" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.14.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.14.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "8c614730a352939325a5aabe8f500212" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.15.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.15.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "5353e7fa40c00e7263ef43bb2f4fe695" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "1df473a2f6133556d9f9e85474e29095" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.16.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.16.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "182aac736f94d983d5ebf291a0c01b9f" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.17.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.17.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "ba77f2654a5ab663a713285c3b204190" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "0d32c154cd89d78ce3c18afc47aefd57" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.18.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.18.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "abf2781904aef5a55c5e5fd2221d7d0d" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.19.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.19.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "8445fedd5ed283b0d49617d71385640a" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "5198be60b3845ab3d673b9442574e2a6" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.2.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.2.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "e02175fb66b315c136bbf38fc1d8534a" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.20.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.20.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "1da008f365f6aee067112f17561fdc9f" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "aa6c57b97c3082d8c3c0d47faa8f099e" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.21.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.21.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "142f7476553a6e3ebdc6beda85a3c494" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.22.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.22.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "efbc9292152b19237c86a6117935ec1e" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "a8c1c802961df641360b798024e069fc" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 31099392, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.23.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.23.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26228736 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 30947328 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 31094784 + } + ], + "md5sum": "e9c106ada593b95a906c123b09b30fb0" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.3.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.3.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "9dad5addda09be73f9dc50216eedf756" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "8d7116e51ca22d4951d71d4302fe8750" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.4.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.4.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "3e1bfbee3e96640b11275ecd608fdb7b" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.5.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.5.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "dda80412253450b195cb107d66f086f8" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "4f9261ba5c285712074029722630a887" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.6.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.6.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "d15fda15d9cad9ff42915665260d9a35" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.7.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.7.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "4f0a8544176537834c7854517f825fd0" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "64237e98a904632f22ef4c6e46dbcee0" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 26233344, + "records": [ + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.8.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.8.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 18929664 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 23648256 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23795712 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 26155008 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 26228736 + } + ], + "md5sum": "edfbe7868ac144cb667ca11516ee3875" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 32859648, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 10616832 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 10948608 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32182272 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32845824 + }, + { + "name": "model.layers.9.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32850432 + }, + { + "name": "model.layers.9.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32855040 + } + ], + "md5sum": "52a4e840e2fd7b42b3aa6b94cf36fcce" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 21233664, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 18432, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 21233664, + "byteOffset": 0 + } + ], + "md5sum": "8e3d5d38ebe7b73cdccf9c8509756906" + }, + { + "dataPath": "params_shard_38.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": "60569958c4cc69a7204a986a102127fc" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 32993280, + "records": [ + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + }, + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 7303680 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 17920512 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 18252288 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18915840 + }, + { + "name": "model.layers.24.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18920448 + }, + { + "name": "model.layers.24.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 18925056 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 18929664 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 21288960 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 21362688 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 2304, + 1152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10616832, + "byteOffset": 21367296 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 2304, + 72 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 331776, + "byteOffset": 31984128 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 18432, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 663552, + "byteOffset": 32315904 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32979456 + }, + { + "name": "model.layers.25.post_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32984064 + }, + { + "name": "model.layers.25.pre_feedforward_layernorm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 32988672 + } + ], + "md5sum": "c5eda88b8a5d8d7e1fdc12cf011933f0" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 7303680, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 4096, + 288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 4096, + 18 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 147456, + "byteOffset": 4718592 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 2304, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 4866048 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 2304, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 73728, + "byteOffset": 7225344 + }, + { + "name": "model.norm.weight", + "shape": [ + 2304 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4608, + "byteOffset": 7299072 + } + ], + "md5sum": "8e4e39996c3e927ff2b1edca7b5e84ac" + } + ] +} \ No newline at end of file