diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3529 @@ +{ + "metadata": { + "ParamSize": 283, + "ParamBytes": 2008559616.0, + "BitsPerParam": 5.001471576767255 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 197001216, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 128256, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 197001216, + "byteOffset": 0 + } + ], + "md5sum": "5d010d48d5a5c425c45c0d7cd4387b3a" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 24631296, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 128256, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 24625152, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 24625152 + } + ], + "md5sum": "656fcd9dd25dce4dc098ad086167f9ea" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "54efa82fbed25ecf0f5b05dadbc2af8a" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "a713baf28adb271b25d755c8f5e0025a" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "00d8e141f8e751064fa003525afd8c57" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "80a4bd1be5b2105e8914919d052212cd" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "8022a70d0efb63f1de126b302c00d7ec" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "74aad4f21aa1a0222a1b247c5a8869bf" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "4db18c6caa763b2e8b9b7e4cbb4b941a" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "0dead7bb279fdd66d1541cd847176453" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "4c85faf1e64c45e4c458bee2f184361f" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "c7ddefb155674d04fca6fe13509576bf" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "eb75eed700b9b82f62f98de05e94165f" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "aa74d5f712dbb88a5d8b5062010115dc" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "32ef1844abfd15dab093166e4c93e743" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "e6d430d62b20c17e107c88d6f6163443" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "1f0cec0b41dc8a0db81238aa9cdde254" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "07cd149e55e526ce31b65ec99dee8842" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "8f7a04a3268c40f83d767e7a828e7751" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "b29b58deae384c71a9f83b21c53b9f7a" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "df15177e1c19acf500cd4495105a5b42" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "0816dfcfdba4b221659beed330c8f3d2" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "408e618a17c7082f268616804cc9beeb" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "87339ffbb75a054bafb56505969b1652" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "aad6dd811a279854b26c2359506f4311" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "b88ff4a4bfb12ff0882cfea312a54b45" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "8f5d85221c94c2362a1d620360e3b8ec" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "bacd4a5379d5fa1e63289a9c74a113dc" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 31463424, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + } + ], + "md5sum": "7dfc35f27ee5404f81892ce7c9a5d12f" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "d49d466d3ebd77d04a218bded748ab54" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 31463424, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3145728 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11010048 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11993088 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16711680 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17307648 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29890560 + } + ], + "md5sum": "504b0f66fc026c6034e2817dc9ab7d44" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "a5a3b173eaf59607cc1e167045a83ce2" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + } + ], + "md5sum": "8e293a4150f32c7722923e679ae26827" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "9a578d4e68d67f2dc0713f30435152eb" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + } + ], + "md5sum": "e74c516539781bdb3fd5f32860b5487c" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "a26966b436622a0d67755285a321c0c8" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + } + ], + "md5sum": "97138a8d9dd85dc69c10db0d2a5d4b5e" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "bb0c7d476083da42c9124d04ecd815bb" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + } + ], + "md5sum": "ff111ffe7408bd99e82274680e5f0cf9" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "9bd90a2be3fd83d91e9c51d63f00f2be" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + } + ], + "md5sum": "6a7dafe93476fd30b3b14d766501c720" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "401e5da19961b59149a3377b88e75834" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + } + ], + "md5sum": "fdbe764d259a4ae4d82c016b3105c0bb" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 31481856, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 0 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 3145728 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 3151872 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 11016192 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 11999232 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 16717824 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17307648 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 17313792 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 29896704 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31469568 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31475712 + } + ], + "md5sum": "1bd628feaeec48637c2ccc67eb0025f3" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "e4d9c12e8882c79bb04bd39b66e8eae1" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "2b22f5888c2171216642b91c064a8cdc" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "4069a9bce66d81a8e1c3ffe20b105aca" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "6e96cde9b821f89c03c5f11ddc071b82" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "082abc333e8eb5a9c6615a19ad8e0b1b" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "b3f858c968b64a027a2917cb2dfd3c73" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "b7509becd1d85f5d7784b21a49e97112" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "e7809aabcdaf7e7474a81164c59e8b10" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "0c86d79d48cbd7ddcaabcb8d4a5f3671" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "f483c1f5336e03885893678fd2a6b672" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "aea995dca12b6baddbff81c5d6f8ab35" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "758f282cfc1b51e72fc4f2338d50c7b9" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 16384, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "9bddff6000b9b0b0383c4b2ab3fc31ed" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 31469568, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 3072, + 1024 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 3072, + 256 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 12582912 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 16384, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 3145728, + "byteOffset": 14155776 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 17301504 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 5120, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7864320, + "byteOffset": 17307648 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 5120, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 983040, + "byteOffset": 25171968 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 3072, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 26155008 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 3072, + 96 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 589824, + "byteOffset": 30873600 + }, + { + "name": "model.norm.weight", + "shape": [ + 3072 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 31463424 + } + ], + "md5sum": "aff47ac2fda02616c1be948328f19991" + } + ] +} \ No newline at end of file