diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,5623 @@ +{ + "metadata": { + "ParamSize": 425, + "ParamBytes": 4719681536.0, + "BitsPerParam": 3.2403624459229405 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 262144000, + "records": [ + { + "name": "lm_head.q_weight", + "shape": [ + 512, + 128000 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 262144000, + "byteOffset": 0 + } + ], + "md5sum": "f24aaa6056ecbff2352411223821b309" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 33366016, + "records": [ + { + "name": "lm_head.q_scale", + "shape": [ + 32, + 128000 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192000, + "byteOffset": 0 + }, + { + "name": "model.layers.36.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 8192000 + }, + { + "name": "model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 8200192 + } + ], + "md5sum": "c42612875deb70a36ee1367c77b10553" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "d0868656246d4fe7cff4bd5884d6daf4" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "d79ccfa535799db26eecf62045d1d08b" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "f9e95aa5063ffe2d3bf51c638dd7b139" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 26370048, + "records": [ + { + "name": "model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 0 + }, + { + "name": "model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 786432 + }, + { + "name": "model.layers.36.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 2359296 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 2367488 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 14950400 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 15343616 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 23732224 + }, + { + "name": "model.layers.37.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + }, + { + "name": "model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 24002560 + }, + { + "name": "model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 24788992 + }, + { + "name": "model.layers.37.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 26361856 + } + ], + "md5sum": "9a968cfe4b4f1578bdaaad0034c0c81e" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "da6d1509f91e9052c6b64d3afbd3ca17" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "f92deafc55d26a8e65f2f5588476cc72" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.37.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.38.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.38.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "824c6cc6b5401036b2daf9696c4034ea" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "a2b00c285227cb1f17b40f47c9d32cd4" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "bcd53c79c167060c4666e56d7a9442b2" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.38.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.39.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.39.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "b8e7d32b24a9e2e8045429809fecd310" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.40.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "228620e958e9020bc26db3b16fb5b551" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.40.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "356056454133aa7a65a53b186e8be788" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.39.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.40.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.40.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.40.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.40.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "cbb374ed85bfd3a57cbde7305ee5df81" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.41.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "594e80b7b9da7c03e83ae9e81f457447" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.41.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "2389fd72c754fe5856237898a9aa3060" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.40.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.40.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.40.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.40.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.41.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.41.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.41.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.41.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "d56109a95b5564c5a4bf2968b3a43ae3" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 262144000, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 128000, + 512 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 262144000, + "byteOffset": 0 + } + ], + "md5sum": "5cb0046e415bebcd14104c5d33a28339" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "60b82c859eb00072e5cfecd731856387" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "be93c66b9073a1c6b5053155db389f84" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 32202752, + "records": [ + { + "name": "model.layers.41.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.41.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.41.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.41.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.norm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 128000, + 32 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192000, + "byteOffset": 21635072 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 29827072 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 29835264 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 30621696 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 32194560 + } + ], + "md5sum": "5f76625d38056ce3ef595b08b3548dd1" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "7d1c82479b08bc86bea471818854ed25" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "d06d42d0ccf470ad9c0894bb52d55107" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "3d0bfc811ff0fbb8067e23eadbee3bd3" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + } + ], + "md5sum": "e9d0bdab741b0ab18a174f8c5d10390c" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "aa8aa6c92292cb03dd0329c99842d211" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "75ee28e586317a92b68071868486935d" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "21337f4376d5ce4614e18b06de582110" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "3a48b23f609c1bcf3ffcf31d67176e1b" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "906795adb932dcf9b5f31965739470b7" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "03a70ecb5d94b57f160b71a88bab86d7" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "5964a661f94059a66148564bf0550ce5" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "6b2268fde1bbb61c2c7957d922d12b9c" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "35c8924776a7b8952552924101672092" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "9d7dc5e1a6a87c3249469aabc9f029b7" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "7c7ba3a87dbe1f4763ab22c2fb801236" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "e1f63f06b17b2aaf99bc1842cae58365" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "ba134edbef6dd31652647eb63ba97074" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "9894fd6688cc187aa71648e8d602b0ae" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "55b68d6a448004a80757b3b80f513f4a" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "b2a827c352064fe9f66e0fc5f6879aab" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "3babd907489aa599715e79bdcc2cb6f6" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "68c95afe32cabd902a66af1ecf6609b2" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "f2946167eb7384712e88143076cae79f" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "812601870bfd928d9aeb203efcded6f2" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "8fe544da2e3f10b7851725586003c28b" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "d21880451647fa1431623091f92569e0" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "1fb6a74335709301f4422787547c3ad8" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "de551d55089196be3fe1ad3e3208a8f5" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "201676c3f48794ec0f8090677bbf1412" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "4949dcc07a4ccad9a297613f27de0a47" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "2d9f6764707e27653d24a840d93517ad" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "7800438dd0bd2eaa7b2828a08a123965" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 26378240, + "records": [ + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 24002560 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 24010752 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 24797184 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 26370048 + } + ], + "md5sum": "17a115872f8d97c7e8349b6c25cb13ef" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "cb97096ffd384b8ec26b574401a9d6e8" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "357160b9165a1e2aec38cb2dd50906b4" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "906a23c93568b61fc5bf871a3467e2aa" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "c20f2aaca668538639af304306fac88b" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "13e185a7dbdc506ea5beab5c5ca1b660" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "da6b8914e3e5d0521657f782a3edb267" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "9e996f7c627ac8ad626145904b13a77e" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "bc5d0c7c49a7b5bf90aa024bcea8fea0" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "1e0664d7102580f1ce81171845934879" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "eceda4e63ae94d3b2dd7f8dba8d5db5a" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "dcb9c96e2e6a56c16e5d7253d810d7e6" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "16d0d602532313d638879049568dec67" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "676de2bb13313916ab8acb1191646f3b" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "c40f6f1cd508735352c6e3379bbe471d" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "94c5b51d675a0cc10fee598d3e331a3c" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "e2906a503e257dc08cdbd1bf15e46b1c" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "a91eb53aaddbe63c69b9eb80f8e63aca" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "3ac3ea6d299c28fa50286cbfbc21d8a5" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "e714db2d47cbaada663e6668d38cad67" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "d9bca83ae0bb51a3d8e8d13e2d446f79" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "f67c9dabcde3c1277f4b8025fa0b90c0" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "a46995e85015d027e05cbbd5745ed187" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "a30888e8807043195628ec4c8c924ae6" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "103456b28c13608658dc746de2a6c747" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "d3312f9a759b5bade7a91e71fd6b6227" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "29d6a7f9f0c360249b5c4301dc309c48" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "644a169d139b9ef354720c970fd5b257" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "d76bbe0b7abf42828af086a3ecada8f3" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "8694d98f9a18fb0fa92c4beae4e66e4f" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "c2c9c70e5fd297e2429d5a24aa9c4206" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "4d9cb56add994bf5a7a997ab80789234" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "d598bf69a0588531dd992bab6b81f3a2" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "2e22a9b4b2d7c6d533f924c15ebc6407" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + } + ], + "md5sum": "a64d473c7648f064258e2074aa0a8242" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "ce7e9783b1f212a8d023fc4b679f4429" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "5af5e6c16815dc064fdb0a6b8f36510b" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "2b729d52852327a8767ca04ed7cd0faf" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "c586da95f83550a2bc557ad6f7cd93d6" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 26378240, + "records": [ + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 12976128 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 12984320 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 13770752 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 15343616 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 15351808 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 23740416 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 24002560 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 24010752 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 24797184 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 26370048 + } + ], + "md5sum": "fd943cc5c7ed2c1e79f84ced4f847580" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "d386547e760a06ad1fafbc6a2ea8a8b8" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "f57868f389b2d7bb61c76bc51f665043" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "4c4fecfc74c23d8c6457a9d4bdf8b16b" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "fa40eb66b8880d0737f03bd70dabe584" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "df208ff582f68f751f8f69bdbadb5816" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "a2c946b4cee44dd84eba94640aaed41d" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "67829fea653e46bd380d334f6b28d786" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "81272f36765c1a8249b55c1bd4645afe" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "167c37963f94af725cabe62b0f5e4c35" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "225fcd59a886f5ac5e6e909300767dc8" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "2b56c29aef76a3f9fc85ced480991090" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "46113190c3807506f2815bbf7cf22633" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "5b04909925bc636aa0b684d9d869312c" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "b506044ea14cddb3951b1cec99c7ac33" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.28.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "ae1247434d26a8a20f251987402fb7ef" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "b026f375987e57fb1ce191fee3eadd7b" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "8278e0adc8077b0a3c4cc1cfb0f86a30" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.29.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "2ccf198b447efb649c3cd884f411fea9" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "a8377f50b8e5eb61e9918686875cbbaa" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "b44fd7a72076e86dfa296faa080c212e" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.30.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "a522b66345ce26d85829e2d8995f1056" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "cb877782104828540268792016988d0e" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "715339be4fd3e72bd09e89a0a7a0d715" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.31.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "c68ec24ec9e3604fc0318dcf587c07fe" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "431575ae4c02ef717193cd6607e1811e" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "0b3f2afc6d45316cd4b8875db1e24ba9" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.32.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.32.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "a9e8ca66227ca7d87c718e53ca225d67" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "27b160af5088f849824236a7a0d6a4ce" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "044cfad26346705bde336c9356ef6d65" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.33.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.33.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "95f532e34ac814f7044bc1c235cf584e" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 25165824, + "records": [ + { + "name": "model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 1536, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25165824, + "byteOffset": 0 + } + ], + "md5sum": "bd40caeb468a325d4c992161bc72a73c" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 50331648, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 512, + 24576 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 50331648, + "byteOffset": 0 + } + ], + "md5sum": "a317be74fb52857a9dd2aa1dbea3783d" + }, + { + "dataPath": "params_shard_126.bin", + "format": "raw-shard", + "nbytes": 24002560, + "records": [ + { + "name": "model.layers.34.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.34.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 21626880 + }, + { + "name": "model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 96, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 21635072 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 32, + 24576 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1572864, + "byteOffset": 22421504 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 8192, + "byteOffset": 23994368 + } + ], + "md5sum": "1992f20cfa253393207fa4fe85acee18" + }, + { + "dataPath": "params_shard_127.bin", + "format": "raw-shard", + "nbytes": 21626880, + "records": [ + { + "name": "model.layers.35.self_attn.qkv_proj.q_weight", + "shape": [ + 512, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 12582912, + "byteOffset": 0 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_scale", + "shape": [ + 32, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 393216, + "byteOffset": 12582912 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 512, + 4096 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 8388608, + "byteOffset": 12976128 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 32, + 4096 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 262144, + "byteOffset": 21364736 + } + ], + "md5sum": "0a29195282fa3c68f4aa8dc8907da369" + } + ] +} \ No newline at end of file