diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,5687 @@ +{ + "metadata": { + "ParamSize": 405, + "ParamBytes": 8136417280.0, + "BitsPerParam": 5.000860280867739 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 81960960, + "records": [ + { + "name": "lm_head.q_weight", + "shape": [ + 32016, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 81960960, + "byteOffset": 0 + } + ], + "md5sum": "945ace250f3148c92ddbd7ea53920d1b" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "b45c53afed318d44eb5bdbcbe8106bfa" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "6138c6434b74edeb7c7bb8a55de3a2c8" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "7a34e43f99ffc16d45c6d1f13643798b" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.31.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "a52117b1780dda885d273e5d108c1b42" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 32896000, + "records": [ + { + "name": "lm_head.q_scale", + "shape": [ + 32016, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10245120, + "byteOffset": 0 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10245120 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 10255360 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14679040 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14689280 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14699520 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19123200 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27970560 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 27980800 + } + ], + "md5sum": "eced6b6d46ef27870139158a95b739e8" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "a38741800dc1d58850a70b8e03fb31b2" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "9cc6b3a8d11d1ea27b36b65feb9f3852" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.32.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "2c84c35240a741847b58caa6aa196aae" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.32.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "f0147a3e80b37772e5eb12fd82b327fc" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "c1295db557bd33278773def6d63c81df" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "f697adb0773e1d709f9b0314855f7aa0" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.33.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "77983c81c2041c70d0bcb8e4b5f9e7f0" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.33.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "c2199698da81e5b16b7282d94645d18b" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "fedc7b035f743a6bc71f204a980cde3a" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "8cfae896ee1e4638db7b1a130bca8971" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.34.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "dce8e376451309ab6bd2cf86788b512c" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.34.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "4ae5c9a6b35331b2e230409d18b68071" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "ef58039eda7467775f7615091a979f0a" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "e000b8acfae3a42f9e92c32568c4894c" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.35.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "4e044759a8a76b49fc6928f78d3353ac" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "b94a2b292fe6240c33b246249dee8fcb" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "37fbb371d638514a3ebaeec17dca6e79" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "d5a051e565f09ec3880dc459b50e7368" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.36.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "1507999fa5d4c03e955804c989ce89a1" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.36.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.36.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "efe2487c9622d5c7d1627a8a14ea5d88" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "b6890470c151bc605fdea6ceee9a3862" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "2c9dcbb10d7c386b974050454e1e33ee" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.37.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "829544538787e2e1302848c5042f95d0" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.37.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.37.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "dd558f724f446877cc5184c3089e8078" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "b84701932044006dbf0b17560f7304f7" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "632f986ab1a10a33b8a011fe05e25173" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.38.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "65fa75442274aea338a0b840009441f3" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.38.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.38.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "ea6b3e81011906c62e6049de7e3ee7bb" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "ecae6513c333cbef0f1734bc7d6c017d" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "8f3b2743e249d733703cffea04d4ed30" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.39.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "2c7062295ecf1d7e8a1437493b224c89" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.layers.39.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 14755840 + }, + { + "name": "model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 19179520 + }, + { + "name": "model.layers.39.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28037120 + } + ], + "md5sum": "4766597919820234291de7127391e419" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 81960960, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 32016, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 81960960, + "byteOffset": 0 + } + ], + "md5sum": "67c37b58828ed4f9f9df8d9463035ac5" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "4f75b3b998ae033bf6164b3df49dd176" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "e1eee5d14901112b9849548a87501cc3" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 29434880, + "records": [ + { + "name": "model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 0 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 13107200 + }, + { + "name": "model.norm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14745600 + }, + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 32016, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10245120, + "byteOffset": 14755840 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 25000960 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 25011200 + } + ], + "md5sum": "dba4bfe1f25ee042f58cd28adc36512f" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "f5147ee54e162e683b60f7ee49d43266" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "cb7162961c05fcabc73b2adbb9171514" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "2b9bd71f4119f9e2d6c356d908e16fb5" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "fb5a3118d7d9aaf1d5ed39b3152c5d02" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "23b463873de510689e88e7402ecf75bc" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "5ae70439495901f4ed68b897d6177201" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "27d744907333cdbb07fe900ea194714c" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "f08d8402610b437421e4c6549dd29001" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "e19065d9771b871468b6328e9313cb6e" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "7c4c16af35dbce78f079cafb92835245" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "08d0088b439a3f36f37dd2c452a3e758" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "ce2b17863400db75daea03113a581e95" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "df4276165352687cd5ed450a6edfcd6e" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "765dfaef58d3787efe19cf83c34bcf95" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "b57495eab7a3b4eaccd91c4f152addc9" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "cff664773cc49d10c445f9ad2bfa187d" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "72cd38116322aee3de379777adf4a9a5" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "da754466213bc205581ac859f0f27b5f" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "a09d818f8225b930a89ff3bda516b3de" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "e020b807a37f266753b27b077d58b091" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "e03dade1b10eb401c4c88df9e18382eb" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "4846309cb98f7b17a0db8d1aaf15509c" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "5cbe11b143887b57467c246e6f5c6da0" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "e461e50c8d101a18b2cb319279334d0c" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "fc36b85842476f05dab6655801188f4a" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "73869ca4a737c41ca5d0f7ecda4199c4" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "2961adb14e3435d29f0ff2a7cf30b933" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 33443840, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 28518400 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 33433600 + } + ], + "md5sum": "f3b06cc00485dd07e9aa53535e5788c6" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "4f5f560187bbd7b374a9ffedfc4b38bb" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "df42c7eb15a2427d1c9ba3be504b8a6c" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "b0ce6d95c2ee277d14ecf42d1f292844" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "4449c093be2af152f7ff0562aea85e4b" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "86a518942f00dbb2bd243c1e8f03a51a" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "139f9e3ce114b00ba2464b71a890726a" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "4441f9921e0fb161a04074e132239659" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "4c1bfccc1f182b40bcd63740c8791a95" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "3ca9695c6bc2a925a4954f83d09e3565" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "6dd8fae210f1ae8728cabfe38112aef8" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "12a019b632797dee7807e4471f3c3fef" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "6c07214b8429524aede490734818e8dd" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "c030775edbf6de36a8840abf2fbeaf15" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "302dc895789ff1bbcd124e706b78fb80" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "9234be422f294def4f7fe17778516af1" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "372576d5dfc814472bea2404d0cadc69" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "2cdec7b2994580dfa6cdc186ffdc183c" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "f3008cd28b578d65d7682313ab6b2577" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "abe0fff292b4d4ac6940b43047e697d3" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "eefe05da7ad2ff42f41d02307f6f660f" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "207fc5198e817011ca598687ea18d93d" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "710ba5024f3f289b40e66bb152224d35" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "60aed1c34329a2d0da2cd8f9ef2b25ff" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "50e4fb5b59a2742169412b53decfbb7a" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "c1a4e2ad43aeb427ec85db3744bf2d11" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "e22265fecf31e96b4aff2db3e122d187" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "5d814f8feeace6bf0876d33a49349e0e" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "545a7bbaaab5cf40434acf63fef5a147" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "ba4448b1aad1e0db94601dbaddd5ed78" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "0ba7296e09372bb637e42c174d44a2a1" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "8685e68adef100a6e68f503e328c5d77" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 18196480 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 31303680 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 32942080 + } + ], + "md5sum": "2565a769ca7a90f91051622fe3a2c8ce" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "98809825700ff5feb6c5780aaae52bfd" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "99afc35f09c1b437c69159d14b0a0e84" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "8ea236b25c6b4f6c3654f0a477499953" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 32460800, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 4423680 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13271040 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13281280 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26388480 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28026880 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28037120 + } + ], + "md5sum": "33b6cbbb519b750ec52722f67d0fcca7" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "18e6e974bd9cf9efb0633b89416b2e67" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "9b4f1853755a9ea54fffeadc2ccae1fe" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "1d918e4638f84eecd96563f2a4ad7dab" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "ba7208ed014c44f86788b3f61f83a866" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "50fe65d9b72a65cce05130cc6b066426" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "c3066ab5873d709b5cc06d5d6244e892" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "88a0e43972fc45b78a40ce72be34b2c6" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "1dde694e5a3d881413a5e890a6bcb734" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "9a71c34e2f7ed8444f8fd02c80abe485" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "2fde0bee89059fb199ce3b8cee61b4c4" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "a246d8a682dd0cb4a42edd65ad92d4eb" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "9c2d80061fc38a7690b9cba1392226ee" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "080407ee2cdd6e8f33aad6f9917df24d" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "a2ca56a336561eb38b860ee7459b7214" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "fda06979211319b7202fde07f55d58e6" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "b19d4263746ab2f5be24c4c54f7916db" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "a0d02e525a9bd579dd25b12c9887f69f" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "16718bd50ae6032c2196b2d0373fc33f" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "a79eefec5608a3ffb675a4646c7dc213" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "db71a7e7fbbcb53953ae565b0d75402a" + }, + { + "dataPath": "params_shard_126.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "003566409c56433b4a9c9608f8277add" + }, + { + "dataPath": "params_shard_127.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "170d631a1c105edb19e401e16923b953" + }, + { + "dataPath": "params_shard_128.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "94b7d53f646d6bef26416e573716838d" + }, + { + "dataPath": "params_shard_129.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "73cc1cec880b4cbba6382f1e91f8080c" + }, + { + "dataPath": "params_shard_130.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "32858b9deabc490d793e8e83ea66c22e" + }, + { + "dataPath": "params_shard_131.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "657a87f31403b36779be5d19a25992be" + }, + { + "dataPath": "params_shard_132.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "e78bde374725b11e23b66d4f55d1c259" + }, + { + "dataPath": "params_shard_133.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "48c85b3f5274b88a22bbc6a316123bcb" + }, + { + "dataPath": "params_shard_134.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "10b30c82c80fae3ffd573b1dbab8e497" + }, + { + "dataPath": "params_shard_135.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "f0286ae5a21eb4f367d8ed9af9b26ca4" + }, + { + "dataPath": "params_shard_136.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "d6e91bd03a9e4f5221328d0b29fd61b0" + }, + { + "dataPath": "params_shard_137.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "7e284e26e9e70a08de2bdf6b8750c019" + }, + { + "dataPath": "params_shard_138.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "ba7f168c33fa91c4765cd57ae4c36d20" + }, + { + "dataPath": "params_shard_139.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "ef4bb44f2411003384447d197963612e" + }, + { + "dataPath": "params_shard_140.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "d20d01205e600365904949f5474b6773" + }, + { + "dataPath": "params_shard_141.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "96a5954ef286ff147f51a24bcc4b7749" + }, + { + "dataPath": "params_shard_142.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "af0b3b1c47a8e97c85d73bcab87869a8" + }, + { + "dataPath": "params_shard_143.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "94b57d075ca62b8ba49182761490f274" + }, + { + "dataPath": "params_shard_144.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "d77d6c1a113ab4a17cff46873f66593a" + }, + { + "dataPath": "params_shard_145.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "ffe7894d9f3d613d1583718507ceb8d8" + }, + { + "dataPath": "params_shard_146.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "38916459458e10b9a37326ae32c065ab" + }, + { + "dataPath": "params_shard_147.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "cbaccb9dae527e85328f2aaf7fb3ff6d" + }, + { + "dataPath": "params_shard_148.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "97944c6bedd2c8a20e4bc5195524ce1a" + }, + { + "dataPath": "params_shard_149.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "1e150ff0ceef7373cd3ff96401d677be" + }, + { + "dataPath": "params_shard_150.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "9519f4e3f4f7f706122ab28956f34713" + }, + { + "dataPath": "params_shard_151.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "1b96187e20bcba1ec06aff148da31b69" + }, + { + "dataPath": "params_shard_152.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "e284c6d3d31de1e084c4fed32b34a690" + }, + { + "dataPath": "params_shard_153.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "d85f6b470869549238fad90ab8a1e214" + }, + { + "dataPath": "params_shard_154.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.28.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "d7358418ec18efaaba6b8561018ea724" + }, + { + "dataPath": "params_shard_155.bin", + "format": "raw-shard", + "nbytes": 35389440, + "records": [ + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 5120, + 1728 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 35389440, + "byteOffset": 0 + } + ], + "md5sum": "4a66df06b518cd9892da939b4b4f2e26" + }, + { + "dataPath": "params_shard_156.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "766f845e2cf66cf9b33b68a877823c25" + }, + { + "dataPath": "params_shard_157.bin", + "format": "raw-shard", + "nbytes": 32952320, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28518400 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 5120, + 432 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4423680, + "byteOffset": 28528640 + } + ], + "md5sum": "dd4a3e0fee190114df6c2fa544f5c082" + }, + { + "dataPath": "params_shard_158.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.29.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "d534744e2a5ddfc2f6372c1c4ee27b8f" + }, + { + "dataPath": "params_shard_159.bin", + "format": "raw-shard", + "nbytes": 70778880, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 27648, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 70778880, + "byteOffset": 0 + } + ], + "md5sum": "226f201efc312f42126cdf06abde1d29" + }, + { + "dataPath": "params_shard_160.bin", + "format": "raw-shard", + "nbytes": 28518400, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 8847360 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8857600 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13772800 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26880000 + } + ], + "md5sum": "6704899fcd38e33a97cda850eda69e33" + }, + { + "dataPath": "params_shard_161.bin", + "format": "raw-shard", + "nbytes": 39321600, + "records": [ + { + "name": "model.layers.30.self_attn.qkv_proj.q_weight", + "shape": [ + 15360, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 39321600, + "byteOffset": 0 + } + ], + "md5sum": "b0963025921e6eaf017a8a5b76f6b282" + }, + { + "dataPath": "params_shard_162.bin", + "format": "raw-shard", + "nbytes": 28508160, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 27648, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 8847360, + "byteOffset": 0 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_scale", + "shape": [ + 15360, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 4915200, + "byteOffset": 8847360 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 5120, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 13107200, + "byteOffset": 13762560 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 5120, + 160 + ], + "dtype": "float32", + "format": "f32-to-bf16", + "nbytes": 1638400, + "byteOffset": 26869760 + } + ], + "md5sum": "329c9d0aeddd480dcadeaf8fbdfa8b81" + } + ] +} \ No newline at end of file