diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3851 @@ +{ + "metadata": { + "ParamSize": 343, + "ParamBytes": 798633984.0, + "BitsPerParam": 4.504832386302102 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "transformer.embd.q_weight", + "shape": [ + 51200, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "bbcff0fa08b86d3c207d0067baf2422e" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 25468928, + "records": [ + { + "name": "transformer.embd.q_scale", + "shape": [ + 51200, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 6553600, + "byteOffset": 0 + }, + { + "name": "transformer.h.0.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 6553600 + }, + { + "name": "transformer.h.0.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 6557696 + }, + { + "name": "transformer.h.0.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 6561792 + }, + { + "name": "transformer.h.0.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 12853248 + }, + { + "name": "transformer.h.0.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 13639680 + }, + { + "name": "transformer.h.0.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 13651968 + }, + { + "name": "transformer.h.0.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 15749120 + }, + { + "name": "transformer.h.0.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 16011264 + }, + { + "name": "transformer.h.0.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 16015360 + }, + { + "name": "transformer.h.0.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 24403968 + }, + { + "name": "transformer.h.0.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 25452544 + } + ], + "md5sum": "21200155d3cfbd8b1ef46a254979035d" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.0.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.0.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.0.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.1.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.1.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.1.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.1.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.1.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.1.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.1.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.1.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.1.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.1.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.1.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "73063ed6cc50fac6cb5805d281e2e2c1" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.1.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.1.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.1.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.2.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.2.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.2.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.2.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.2.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.2.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.2.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.2.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.2.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.2.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.2.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "e0b6e0e09d5711ce97be27390531eb9d" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.2.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.2.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.2.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.3.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.3.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.3.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.3.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.3.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.3.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.3.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.3.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.3.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.3.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.3.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "0b30370e6efe5205bb84ad315d66dd89" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.3.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.3.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.3.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.4.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.4.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.4.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.4.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.4.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.4.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.4.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.4.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.4.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.4.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.4.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "bcd0448dd1f620fb21898d553f124b4e" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.4.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.4.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.4.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.5.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.5.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.5.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.5.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.5.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.5.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.5.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.5.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.5.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.5.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.5.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "40831558a9eb61985fb59217dccf418d" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.5.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.5.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.5.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.6.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.6.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.6.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.6.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.6.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.6.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.6.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.6.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.6.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.6.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.6.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "2bce86dae74f79860f155c52bcb9f361" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.6.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.6.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.6.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.7.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.7.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.7.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.7.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.7.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.7.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.7.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.7.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.7.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.7.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.7.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "3b93cb72b930984c98bd02aaeabc6e73" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.7.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.7.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.7.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.8.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.8.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.8.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.8.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.8.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.8.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.8.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.8.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.8.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.8.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.8.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "8c06bea27b95b8031c576d9c8df2cec0" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.8.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.8.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.8.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.9.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.9.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.9.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.9.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.9.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.9.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.9.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.9.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.9.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.9.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.9.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "161c60ecd4fc0b0ee49c90f60be5248a" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.9.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.9.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.9.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.10.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.10.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.10.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.10.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.10.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.10.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.10.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.10.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.10.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.10.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.10.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "0ce4cea7a4c219df61fa52d922cc714a" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.10.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.10.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.10.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.11.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.11.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.11.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.11.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.11.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.11.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.11.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.11.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.11.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.11.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.11.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "3e5d2f55f35058805f6b4a0ab0b3b33e" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.11.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.11.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.11.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.12.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.12.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.12.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.12.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.12.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.12.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.12.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.12.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.12.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.12.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.12.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "a58f5e56ec3d3047a2f20910833a80bf" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.12.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.12.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.12.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.13.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.13.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.13.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.13.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.13.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.13.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.13.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.13.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.13.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.13.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.13.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "e53dd27272d22ad8746d7073f2c0db75" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.13.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.13.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.13.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.14.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.14.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.14.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.14.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.14.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.14.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.14.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.14.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.14.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.14.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.14.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "408b5639110fd4f487b9736ae163303c" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.14.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.14.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.14.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.15.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.15.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.15.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.15.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.15.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.15.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.15.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.15.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.15.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.15.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.15.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "bcb5adb8d28f3090005463ecae311df7" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.15.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.15.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.15.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.16.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.16.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.16.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.16.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.16.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.16.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.16.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.16.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.16.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.16.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.16.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "fabbb1bd044446a45e2b3daf5b81244d" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.16.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.16.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.16.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.17.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.17.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.17.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.17.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.17.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.17.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.17.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.17.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.17.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.17.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.17.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "230582bc0fe650cff3b88b5a363b4dee" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.17.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.17.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.17.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.18.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.18.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.18.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.18.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.18.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.18.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.18.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.18.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.18.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.18.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.18.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "6538665f514523c492c5037fe2eb092d" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.18.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.18.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.18.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.19.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.19.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.19.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.19.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.19.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.19.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.19.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.19.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.19.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.19.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.19.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "2e3b2187339c9cbf714218f9d3a2caf5" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.19.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.19.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.19.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.20.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.20.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.20.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.20.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.20.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.20.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.20.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.20.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.20.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.20.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.20.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "0ec0822d742f58f4d19f2a39460c1c71" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.20.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.20.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.20.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.21.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.21.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.21.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.21.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.21.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.21.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.21.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.21.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.21.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.21.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.21.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "015b6bea73e884208bb568bffedc3f9f" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.21.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.21.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.21.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.22.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.22.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.22.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.22.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.22.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.22.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.22.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.22.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.22.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.22.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.22.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "f1ca94839099428d42046796a91b1f37" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 28356608, + "records": [ + { + "name": "transformer.h.22.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.22.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.22.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "transformer.h.23.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "transformer.h.23.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "transformer.h.23.mixer.Wqkv.q_weight", + "shape": [ + 6144, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 6291456, + "byteOffset": 9449472 + }, + { + "name": "transformer.h.23.mixer.Wqkv.q_scale", + "shape": [ + 6144, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 786432, + "byteOffset": 15740928 + }, + { + "name": "transformer.h.23.mixer.Wqkv.bias", + "shape": [ + 6144 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 12288, + "byteOffset": 16527360 + }, + { + "name": "transformer.h.23.mixer.out_proj.q_weight", + "shape": [ + 2048, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 2097152, + "byteOffset": 16539648 + }, + { + "name": "transformer.h.23.mixer.out_proj.q_scale", + "shape": [ + 2048, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 262144, + "byteOffset": 18636800 + }, + { + "name": "transformer.h.23.mixer.out_proj.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 18898944 + }, + { + "name": "transformer.h.23.mlp.fc1.q_weight", + "shape": [ + 8192, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 18903040 + }, + { + "name": "transformer.h.23.mlp.fc1.q_scale", + "shape": [ + 8192, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 27291648 + }, + { + "name": "transformer.h.23.mlp.fc1.bias", + "shape": [ + 8192 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 16384, + "byteOffset": 28340224 + } + ], + "md5sum": "3f30bffc08d16b8936298c07da47632f" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 52428800, + "records": [ + { + "name": "lm_head.linear.q_weight", + "shape": [ + 51200, + 256 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 52428800, + "byteOffset": 0 + } + ], + "md5sum": "b5c7458234d726812e45275561d6d3f1" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 16105472, + "records": [ + { + "name": "transformer.h.23.mlp.fc2.q_weight", + "shape": [ + 2048, + 1024 + ], + "dtype": "uint32", + "format": "raw", + "nbytes": 8388608, + "byteOffset": 0 + }, + { + "name": "transformer.h.23.mlp.fc2.q_scale", + "shape": [ + 2048, + 256 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 1048576, + "byteOffset": 8388608 + }, + { + "name": "transformer.h.23.mlp.fc2.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9437184 + }, + { + "name": "lm_head.ln.weight", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9441280 + }, + { + "name": "lm_head.ln.bias", + "shape": [ + 2048 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 4096, + "byteOffset": 9445376 + }, + { + "name": "lm_head.linear.q_scale", + "shape": [ + 51200, + 64 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 6553600, + "byteOffset": 9449472 + }, + { + "name": "lm_head.linear.bias", + "shape": [ + 51200 + ], + "dtype": "float16", + "format": "raw", + "nbytes": 102400, + "byteOffset": 16003072 + } + ], + "md5sum": "693188db1d3398a7923eb9a4ec2c9660" + } + ] +} \ No newline at end of file