diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3473 @@ +{ + "metadata": { + "ParamSize": 283, + "ParamBytes": 1449257472.0, + "BitsPerParam": 3.6087652046199286 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 158011392, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 128256, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 158011392, + "byteOffset": 0 + } + ], + "md5sum": "4ebb7070166ba4f81537937d4eaa84b6" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "9992387df8d7361a415b8a8c5335584b" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 31093248, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 128256, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 19751424, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 19751424 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 19757568 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 29833728 + } + ], + "md5sum": "99d67040848f3d6639f0ee9254afe458" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "a9495f936198708cd310e25b639875a4" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 27754496, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13883392 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 13889536 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 23965696 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 25225216 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27748352 + } + ], + "md5sum": "4457c5a71756454bdc429ea63bd58459" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "f68f2e7676fb59cdba8587fc71388e90" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "c0aeae3a55976e73980cfa60c1cc43e6" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "7445d816f526ce49873a9eebb20bf8d4" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 29489152, + "records": [ + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 15599616 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18122752 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 18128896 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 24436736 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 25225216 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 29009920 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29483008 + } + ], + "md5sum": "a0a06d70a385ae44a8fc1c7582b00e1f" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 31520768, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 10076160 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 11335680 + } + ], + "md5sum": "5dfc99fee1a8c8316fff95584e089881" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "d54e8ab1857c8ba9a68be5a726e4abe7" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 27754496, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13883392 + }, + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 13889536 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 23965696 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 25225216 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27748352 + } + ], + "md5sum": "cdf148379035b481f523d874daffb76f" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "399e9d14fcf6a55991c971d99cfce9cf" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "341e8fd26e6903cf7644c01157359507" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "5baa57ad18cea49db36adeff2d703a6b" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 29489152, + "records": [ + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 15599616 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18122752 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 18128896 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 24436736 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 25225216 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 29009920 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29483008 + } + ], + "md5sum": "ecf1a403171a2dc52c7dc7689f403652" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 31520768, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 10076160 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 11335680 + } + ], + "md5sum": "1380a934999f0b1ecbe5520cced960df" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "e110c8367796e04901d7c958c5dc091e" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 27754496, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13883392 + }, + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 13889536 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 23965696 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 25225216 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27748352 + } + ], + "md5sum": "21018a49eaaefd5275f821f2d5120ac2" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "cc4f701f971129f4e43173638a079ef6" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "9f7a13d061f6654e1fbf6f48d5bc7b42" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "4a97bbf94a2c241579048a25ad3c2ef2" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 29489152, + "records": [ + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 15599616 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18122752 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 18128896 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 24436736 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 25225216 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 29009920 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29483008 + } + ], + "md5sum": "b170c7ea6610cddbb31ca7c1a36bae7b" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 31520768, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 10076160 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 11335680 + } + ], + "md5sum": "a8528df444842cfaf6f950fcbebe222c" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "b66f53151d306923a4b8235e3b10aac2" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 27766784, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 13883392 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 16406528 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 22714368 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 23502848 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 27287552 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27760640 + } + ], + "md5sum": "3df675eb984fff526532fc68953d24ac" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 31520768, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 0 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 10076160 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 11335680 + } + ], + "md5sum": "862a10536d33cdc5a3cc477b9618bff4" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "0d62826670dd47f736bbe87e4fdb8450" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 27754496, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13883392 + }, + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 13889536 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 23965696 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 25225216 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27748352 + } + ], + "md5sum": "35732edb4e7ec72f3b4388d61c0dca7f" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "afc170578cefa3598f0dc43cf10b315f" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "e6588749cec1e80ec01e462965c7423a" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "9ecf98a995d955fdb9284721f280cd10" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 29489152, + "records": [ + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 15599616 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18122752 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 18128896 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 24436736 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 25225216 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 29009920 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29483008 + } + ], + "md5sum": "5f25ebd0b5b1940a6d9b9c324ae9750d" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 31520768, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 10076160 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 11335680 + } + ], + "md5sum": "ddce323a0917881c24baa8f0b6fcb547" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "3bd4b2f049b359ad4c4944035502f2d0" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 27754496, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13883392 + }, + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 13889536 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 23965696 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 25225216 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27748352 + } + ], + "md5sum": "31c1a2758579ea7870bac5572206d3ac" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "906d049d30afea2c1bc6617049b7ec2e" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "ed3ba515862aeb7a4ff9d1f7ad56683a" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "b7e879b4f8659cf26dd69f0fc5bbd6ae" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 29476864, + "records": [ + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 15599616 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 15605760 + }, + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 15611904 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 25688064 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 26947584 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29470720 + } + ], + "md5sum": "f640659d88532d7669f9fec9d4295997" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "9006edf6e7df22144da01979593879c8" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "11718eb29d38b89c3a4e94a3d9eb20cf" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "ce845272ea98e61a2321ebb801e052ff" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 29489152, + "records": [ + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 15599616 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18122752 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 18128896 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 24436736 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 25225216 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 29009920 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29483008 + } + ], + "md5sum": "33ef53d5203d62692748890a92168f74" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 31520768, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 0 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 10076160 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 11335680 + } + ], + "md5sum": "2f15f7c89cc6516448c54be7e3f06e0e" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "6ac142f8d7e1aa980f2c5cecbeeb4fd3" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 27754496, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 0 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 2523136 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 2529280 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 8837120 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 9625600 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 13410304 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 13883392 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 13889536 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 23965696 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 25225216 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 27748352 + } + ], + "md5sum": "6c33923a012d709d9825a5e96b891a3e" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "3ef34e8fcc38ad4254e2e24c74d3562c" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 32321536, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 6307840 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 7096320 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 10881024 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 11354112 + }, + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 11360256 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 21436416 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 22695936 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 25219072 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 25225216 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 31533056 + } + ], + "md5sum": "1042d5ae6f652e3c66934a00bcae16f9" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 20185088, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 20185088, + "byteOffset": 0 + } + ], + "md5sum": "8e535bb8775b484ef8e150400cec2b56" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 29489152, + "records": [ + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 0 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 3784704 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 4257792 + }, + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 3072, + 820 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10076160, + "byteOffset": 4263936 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 3072, + 205 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1259520, + "byteOffset": 14340096 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2523136, + "byteOffset": 15599616 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 18122752 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6307840, + "byteOffset": 18128896 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 788480, + "byteOffset": 24436736 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 308 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 3784704, + "byteOffset": 25225216 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 77 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 473088, + "byteOffset": 29009920 + }, + { + "name": "model.norm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 29483008 + } + ], + "md5sum": "571811844c2e34525275a9e72c7bb56d" + } + ] +} \ No newline at end of file