diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,4737 @@ +{ + "metadata": { + "ParamSize": 399, + "ParamBytes": 1591545856.0, + "BitsPerParam": 4.1259299471863 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 155582464, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 151936, + 256 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 155582464, + "byteOffset": 0 + } + ], + "md5sum": "d5cc47184be6ec4ae9ee0f46e114d841" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "d99917bbea3dc2e3aa2afd418899ad37" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 33346560, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 151936, + 16 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4861952, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 4861952 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 4866048 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 16138240 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 16490496 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17195008 + }, + { + "name": "model.layers.0.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 17199104 + }, + { + "name": "model.layers.0.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 17204224 + }, + { + "name": "model.layers.0.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 19825664 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 19907584 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 22004736 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22070272 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 22074368 + } + ], + "md5sum": "32862b036fc125b13ea8a96ec68a00b3" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 28480512, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 352256 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 22896640 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 23601152 + }, + { + "name": "model.layers.1.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 23605248 + }, + { + "name": "model.layers.1.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 23610368 + }, + { + "name": "model.layers.1.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 26231808 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26313728 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 28410880 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28476416 + } + ], + "md5sum": "fda38aa7f9279ea1e27875c71f721b81" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "d76f77d16aa943681ba80db8c2e10ad1" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "a219743dc340222a56ebae3d752a5022" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 32249856, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 0 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 11272192 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 11624448 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12328960 + }, + { + "name": "model.layers.10.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 12333056 + }, + { + "name": "model.layers.10.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 12338176 + }, + { + "name": "model.layers.10.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 14959616 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 15041536 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 17138688 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17204224 + }, + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 17208320 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 28480512 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28832768 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29537280 + }, + { + "name": "model.layers.11.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29541376 + }, + { + "name": "model.layers.11.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29546496 + }, + { + "name": "model.layers.11.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 32167936 + } + ], + "md5sum": "16016b1c1c82f1893456a221972bea2d" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "73036cd745ec12f737b1537334a68107" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "de8166ef43505e2c542a95bae2d3fd26" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 31713280, + "records": [ + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + }, + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 2166784 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 13438976 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 13791232 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14495744 + }, + { + "name": "model.layers.12.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 14499840 + }, + { + "name": "model.layers.12.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 14504960 + }, + { + "name": "model.layers.12.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 17126400 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 17208320 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 19305472 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19371008 + }, + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 19375104 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 30647296 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 30999552 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 31704064 + }, + { + "name": "model.layers.13.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 31708160 + } + ], + "md5sum": "90643debbc2095b66f1cdf586151b98f" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "132674686d45ec2a96367ea44a74cefb" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33350656, + "records": [ + { + "name": "model.layers.13.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 2621440 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 2703360 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 4800512 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 4866048 + }, + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 4870144 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 16142336 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 16494592 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17199104 + }, + { + "name": "model.layers.14.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 17203200 + }, + { + "name": "model.layers.14.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 17208320 + }, + { + "name": "model.layers.14.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 19829760 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 19911680 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 22008832 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22074368 + }, + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 22078464 + } + ], + "md5sum": "97952458edd8b93a42e4fe4d2c903080" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 28480512, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 352256 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 22896640 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 23601152 + }, + { + "name": "model.layers.15.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 23605248 + }, + { + "name": "model.layers.15.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 23610368 + }, + { + "name": "model.layers.15.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 26231808 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26313728 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 28410880 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28476416 + } + ], + "md5sum": "4cf7ae7da45e105cf668c14b432904e8" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "309b1e2654ad5ac87a71ba9db7a531b3" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "6d86c0e01f1c393b693a360dc681a17c" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 32249856, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 0 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 11272192 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 11624448 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12328960 + }, + { + "name": "model.layers.16.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 12333056 + }, + { + "name": "model.layers.16.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 12338176 + }, + { + "name": "model.layers.16.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 14959616 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 15041536 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 17138688 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17204224 + }, + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 17208320 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 28480512 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28832768 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29537280 + }, + { + "name": "model.layers.17.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29541376 + }, + { + "name": "model.layers.17.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29546496 + }, + { + "name": "model.layers.17.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 32167936 + } + ], + "md5sum": "a7614c6d9b9fd71260341f0027cb5924" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "0e63208ef70ab15d417fe39e133711a5" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "4c150486221403f563305ad2b35ec8ee" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 31713280, + "records": [ + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + }, + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 2166784 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 13438976 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 13791232 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14495744 + }, + { + "name": "model.layers.18.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 14499840 + }, + { + "name": "model.layers.18.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 14504960 + }, + { + "name": "model.layers.18.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 17126400 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 17208320 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 19305472 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19371008 + }, + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 19375104 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 30647296 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 30999552 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 31704064 + }, + { + "name": "model.layers.19.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 31708160 + } + ], + "md5sum": "fb4753cf9dc56d961369c5cec2c985de" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "a1595c9fc7d0ce8042dce3700b0c1ad0" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 33350656, + "records": [ + { + "name": "model.layers.19.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 0 + }, + { + "name": "model.layers.19.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 2621440 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 2703360 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 4800512 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 4866048 + }, + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 4870144 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 16142336 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 16494592 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17199104 + }, + { + "name": "model.layers.2.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 17203200 + }, + { + "name": "model.layers.2.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 17208320 + }, + { + "name": "model.layers.2.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 19829760 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 19911680 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 22008832 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22074368 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 22078464 + } + ], + "md5sum": "544bbeabe238a2b57514d557128a8391" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "7cf7b9215bea3ed714c95112cd030dec" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 31897600, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 0 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 352256 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 22896640 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 23601152 + }, + { + "name": "model.layers.20.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 23605248 + }, + { + "name": "model.layers.20.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 23610368 + }, + { + "name": "model.layers.20.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 26231808 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26313728 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 28410880 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28476416 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28480512 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29185024 + }, + { + "name": "model.layers.21.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29189120 + }, + { + "name": "model.layers.21.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29194240 + }, + { + "name": "model.layers.21.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 31815680 + } + ], + "md5sum": "9501045755506ef1f6065c2603ed075d" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "afe3ad74c03de623a7b315ba8efc42ab" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "fa4aec55ca327b67b45db3a9cf2ab52f" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 31713280, + "records": [ + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + }, + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 2166784 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 13438976 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 13791232 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14495744 + }, + { + "name": "model.layers.3.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 14499840 + }, + { + "name": "model.layers.3.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 14504960 + }, + { + "name": "model.layers.3.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 17126400 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 17208320 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 19305472 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19371008 + }, + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 19375104 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 30647296 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 30999552 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 31704064 + }, + { + "name": "model.layers.4.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 31708160 + } + ], + "md5sum": "233c9f7c42fe80e793b02de9f110f76d" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "ffa9ba2256ec51123fedf9bef3268ac0" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 33350656, + "records": [ + { + "name": "model.layers.4.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 2621440 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 2703360 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 4800512 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 4866048 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 4870144 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 16142336 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 16494592 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17199104 + }, + { + "name": "model.layers.5.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 17203200 + }, + { + "name": "model.layers.5.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 17208320 + }, + { + "name": "model.layers.5.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 19829760 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 19911680 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 22008832 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22074368 + }, + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 22078464 + } + ], + "md5sum": "74226120a39a6f9a81eade3e3fdd6e47" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 28480512, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 352256 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 22896640 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 23601152 + }, + { + "name": "model.layers.6.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 23605248 + }, + { + "name": "model.layers.6.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 23610368 + }, + { + "name": "model.layers.6.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 26231808 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26313728 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 28410880 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28476416 + } + ], + "md5sum": "f7eaf05547b4f9268efc4889fc461013" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "105da4332b5791651fe8ad23d34e0703" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "4885910f9c89e71d187c1da30ad02eba" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 32249856, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 11272192 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 11624448 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12328960 + }, + { + "name": "model.layers.7.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 12333056 + }, + { + "name": "model.layers.7.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 12338176 + }, + { + "name": "model.layers.7.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 14959616 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 15041536 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 17138688 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17204224 + }, + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 17208320 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 28480512 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28832768 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29537280 + }, + { + "name": "model.layers.8.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29541376 + }, + { + "name": "model.layers.8.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29546496 + }, + { + "name": "model.layers.8.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 32167936 + } + ], + "md5sum": "b08271bacb22acd852bab80c696cf7e4" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "167ef84205690c370356cbbcbd02807d" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 30999552, + "records": [ + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 2166784 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 13438976 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 13791232 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14495744 + }, + { + "name": "model.layers.9.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 14499840 + }, + { + "name": "model.layers.9.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 14504960 + }, + { + "name": "model.layers.9.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 17126400 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 17208320 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 19305472 + }, + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 19371008 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 30643200 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 30995456 + } + ], + "md5sum": "8f3db813a02fa77cb09232c475150f94" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "c1791b903e7d433a10cd4188c747c9f5" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "89fea3b00cd92e191c5aa012b1ddc1dc" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 32249856, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 11272192 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 11624448 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12328960 + }, + { + "name": "model.layers.22.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 12333056 + }, + { + "name": "model.layers.22.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 12338176 + }, + { + "name": "model.layers.22.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 14959616 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 15041536 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 17138688 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17204224 + }, + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 17208320 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 28480512 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28832768 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29537280 + }, + { + "name": "model.layers.23.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29541376 + }, + { + "name": "model.layers.23.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29546496 + }, + { + "name": "model.layers.23.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 32167936 + } + ], + "md5sum": "879aea9514515c45c1f1ebb49d78cae7" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "48583652b63403763a46b74b977bcf90" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "9b2e492f2d63c6beea1a9b05efec8175" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 31713280, + "records": [ + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + }, + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 2166784 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 13438976 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 13791232 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14495744 + }, + { + "name": "model.layers.24.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 14499840 + }, + { + "name": "model.layers.24.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 14504960 + }, + { + "name": "model.layers.24.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 17126400 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 17208320 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 19305472 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19371008 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 19375104 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 30647296 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 30999552 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 31704064 + }, + { + "name": "model.layers.25.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 31708160 + } + ], + "md5sum": "3cb95f6560bd760c1a2a758896d5384b" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "617277ac73088087ed3f3145735bc543" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 33350656, + "records": [ + { + "name": "model.layers.25.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 2621440 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 2703360 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 4800512 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 4866048 + }, + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 4870144 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 16142336 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 16494592 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17199104 + }, + { + "name": "model.layers.26.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 17203200 + }, + { + "name": "model.layers.26.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 17208320 + }, + { + "name": "model.layers.26.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 19829760 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 19911680 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 22008832 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22074368 + }, + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 22078464 + } + ], + "md5sum": "31a16a3e3fccc5ffa4030a1a2c631702" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 28480512, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 0 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 352256 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 22896640 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 23601152 + }, + { + "name": "model.layers.27.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 23605248 + }, + { + "name": "model.layers.27.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 23610368 + }, + { + "name": "model.layers.27.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 26231808 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26313728 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 28410880 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28476416 + } + ], + "md5sum": "b43fb10f5c826eecdf46f68d30987854" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "e38a462ca60a210907aa81698b2704b4" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "c4630bddb45c7dbad3c1b2a2aba7f8f5" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 32249856, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 0 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 11272192 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 11624448 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12328960 + }, + { + "name": "model.layers.28.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 12333056 + }, + { + "name": "model.layers.28.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 12338176 + }, + { + "name": "model.layers.28.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 14959616 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 15041536 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 17138688 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17204224 + }, + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 17208320 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 28480512 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28832768 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29537280 + }, + { + "name": "model.layers.29.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29541376 + }, + { + "name": "model.layers.29.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29546496 + }, + { + "name": "model.layers.29.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 32167936 + } + ], + "md5sum": "1cfcb5e43225c9a33248f4b97cfcf87d" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "75e40bcae42e9942bb81d4e2d7500ae7" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "2f104c6512c64038d0727a0d2cc2328e" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 31713280, + "records": [ + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + }, + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 2166784 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 13438976 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 13791232 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 14495744 + }, + { + "name": "model.layers.30.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 14499840 + }, + { + "name": "model.layers.30.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 14504960 + }, + { + "name": "model.layers.30.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 17126400 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 17208320 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 19305472 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 19371008 + }, + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 19375104 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 30647296 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 30999552 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 31704064 + }, + { + "name": "model.layers.31.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 31708160 + } + ], + "md5sum": "40947c61a1c08b54585be319fab5a585" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "5e4ecf1ef60264bf935d9ae3b9ddce43" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 33350656, + "records": [ + { + "name": "model.layers.31.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 0 + }, + { + "name": "model.layers.31.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 2621440 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 2703360 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 4800512 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 4866048 + }, + { + "name": "model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 4870144 + }, + { + "name": "model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 16142336 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 16494592 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17199104 + }, + { + "name": "model.layers.32.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 17203200 + }, + { + "name": "model.layers.32.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 17208320 + }, + { + "name": "model.layers.32.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 19829760 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 19911680 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 22008832 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 22074368 + }, + { + "name": "model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 22078464 + } + ], + "md5sum": "0510bb7da331a8b1e366942b40ce1f03" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 28480512, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 0 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 352256 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 22896640 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 23601152 + }, + { + "name": "model.layers.33.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 23605248 + }, + { + "name": "model.layers.33.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 23610368 + }, + { + "name": "model.layers.33.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 26231808 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 26313728 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 28410880 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 28476416 + } + ], + "md5sum": "a9c7e10570cc96c272caa8d8f7274cb1" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "05befe42ba07b672955d3611c64c15a4" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 22544384, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 256, + 22016 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 22544384, + "byteOffset": 0 + } + ], + "md5sum": "be10bcdeda59ac55a24a3a9fbe8184d0" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 32249856, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 0 + }, + { + "name": "model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 11272192 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 11624448 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 12328960 + }, + { + "name": "model.layers.34.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 12333056 + }, + { + "name": "model.layers.34.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 12338176 + }, + { + "name": "model.layers.34.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 14959616 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 15041536 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 17138688 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 17204224 + }, + { + "name": "model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 1376, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 11272192, + "byteOffset": 17208320 + }, + { + "name": "model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 86, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 352256, + "byteOffset": 28480512 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 16, + 22016 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 704512, + "byteOffset": 28832768 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 29537280 + }, + { + "name": "model.layers.35.self_attn.c_attn.bias", + "shape": [ + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 5120, + "byteOffset": 29541376 + }, + { + "name": "model.layers.35.self_attn.c_attn.q_weight", + "shape": [ + 256, + 2560 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2621440, + "byteOffset": 29546496 + }, + { + "name": "model.layers.35.self_attn.c_attn.q_scale", + "shape": [ + 16, + 2560 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 81920, + "byteOffset": 32167936 + } + ], + "md5sum": "1d5070c06be6f071161a89d1a901f524" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 2166784, + "records": [ + { + "name": "model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 256, + 2048 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2097152, + "byteOffset": 0 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 16, + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 65536, + "byteOffset": 2097152 + }, + { + "name": "model.norm.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4096, + "byteOffset": 2162688 + } + ], + "md5sum": "cf2799fa117c67d0518616f52cdc62b0" + } + ] +} \ No newline at end of file