diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,5367 @@ +{ + "metadata": { + "ParamSize": 405, + "ParamBytes": 6315878400.0, + "BitsPerParam": 3.3887388858384644 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 335544320, + "records": [ + { + "name": "lm_head.q_weight", + "shape": [ + 640, + 131072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 335544320, + "byteOffset": 0 + } + ], + "md5sum": "1a437db123f6f69e92e71c397a55d3b4" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "5eb177591e732672a7836a41f5577df2" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "1ba87d9c05b789023d4502ca960edd43" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "b1372fc49bcb6ba8af821df8adcca88f" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c80a8055454424642314a61b81a2ba31" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 33136640, + "records": [ + { + "name": "lm_head.q_scale", + "shape": [ + 40, + 131072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 0 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10485760 + }, + { + "name": "model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 10496000 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 11642880 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13936640 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 13946880 + }, + { + "name": "model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 13957120 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 15104000 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 17397760 + }, + { + "name": "model.layers.34.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 17408000 + } + ], + "md5sum": "a6659f76bb18925b4bf9d53df440cfc0" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "37648f4a6f1f226a730a410fa27db34f" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "714706947cd2d77803d9ee5148ffc6af" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 30986240, + "records": [ + { + "name": "model.layers.34.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 0 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 491520 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 10977280 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 11304960 + }, + { + "name": "model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 11315200 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 12462080 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14755840 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 14766080 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 30494720 + } + ], + "md5sum": "4ae73f09ddade5071811a640aa6a64f2" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "f4c8e328a47faf07b4de68ad34a4a4fe" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "0afcae721ce2885dc0e3666ce07d49f5" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 0 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 10485760 + }, + { + "name": "model.layers.36.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10813440 + }, + { + "name": "model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 10823680 + }, + { + "name": "model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 11970560 + }, + { + "name": "model.layers.36.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14264320 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 14274560 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 30003200 + } + ], + "md5sum": "6430e33334a9154a49c1438a88121300" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "e728d1976d6b0849d4123c710ca8f651" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a806a25d8350bd77706f259b614e5699" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 0 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 10485760 + }, + { + "name": "model.layers.37.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10813440 + }, + { + "name": "model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 10823680 + }, + { + "name": "model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 11970560 + }, + { + "name": "model.layers.37.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14264320 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 14274560 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 30003200 + } + ], + "md5sum": "a604861ca99a760bf782fb6b1980e7cb" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "6f7c2e3b7a2a7c3b77ea84659d9cd84f" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a40e385130d78c0374d698d7a538541c" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 0 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 10485760 + }, + { + "name": "model.layers.38.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10813440 + }, + { + "name": "model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 10823680 + }, + { + "name": "model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 11970560 + }, + { + "name": "model.layers.38.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14264320 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 14274560 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 30003200 + } + ], + "md5sum": "4ad1961b563742831dd4539abe82b88c" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "6f8511c7299685ad0fb8442facd49ed9" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "6219a0ffa503171c24e3a96e46da778b" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 0 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 10485760 + }, + { + "name": "model.layers.39.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10813440 + }, + { + "name": "model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 10823680 + }, + { + "name": "model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 11970560 + }, + { + "name": "model.layers.39.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 14264320 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 14274560 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 30003200 + } + ], + "md5sum": "0f7042b5566da983d4a0085f457a643e" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 335544320, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 131072, + 640 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 335544320, + "byteOffset": 0 + } + ], + "md5sum": "1dfcc5bbc40546773953b1685cfbae19" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "c8b0d81d49d4e98ed1264f61e7c716da" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "4cd0b394ecc3e774852929773ceed4d2" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 24770560, + "records": [ + { + "name": "model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 0 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 10485760 + }, + { + "name": "model.norm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 10813440 + }, + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 131072, + 40 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 10823680 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 21309440 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 21319680 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 22466560 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 24760320 + } + ], + "md5sum": "dcfa423535d9dfa1421f75050962efcc" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "2fc1a77335e05a3ea535fd9f200fc541" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "2d370f654ecc5fdedeb197662b288175" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "d925e3b7b9acc70c3b7a4e48c6dafedc" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "a8c7488c2be46e0a4bb70c0d65d7ba72" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "566d7baa460f46e04905f4aadd63acc6" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "c2ab76ebf5771f0618b2b18e8fc39bb1" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "437517ab6f356633f9a73555ba9a8671" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "a8b305d845669958aa0a371b864865ee" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "8e4c964e991f53772e59e9f6db6eb700" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "adea2272905532aa5e8cf81c6ff42fee" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "977304437c54f195b5aafe2066a46e25" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "fbc7401e5b83ef6ae1fa19d6f1f11e1e" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "228f87f3328f5c36731c6198bd431298" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c8af750cd5b19b0a76cc1f36b239c1de" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "4138787e1edf23ca7b0d7c418a23d3c6" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "2960f70adf032acc0b64f832928f691d" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 29327360, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 27033600 + } + ], + "md5sum": "f7a73c66a006c7a1fd78e2a687145373" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "4803ac5a5a480f8fdbebe9f84baa9903" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "cadf01e1e611307915899d1ed51df744" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "d19deadff0c78220b33d7c7ee1078301" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "0cb707cf82d428f4a93e976e585e7ef9" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "28804f2960898902ec375cc9cd520d6d" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "87914dae2b9e6124ff0374af46c042af" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "03cec42ce221a48062f225841e7e0b97" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "0231c12d9d938c54a07cc32a46206188" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "0a18953c5e3a0c04241c3fde80879cf0" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "eff3058feeb6a2261b4d3e1454c2d80b" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "1f405aac798c76521f228f248210c4d5" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "a17273257e25708506dfb62d280c4705" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "0f91d3c2d956625e6817783347383b37" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "0c76eabefb64cc0b2cd51fcebd939d4a" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "83ca3774f4da4a8a51a901ba687fc651" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "f0fc7e8df340bec4209125e426db0f0e" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 29327360, + "records": [ + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 27033600 + } + ], + "md5sum": "16b24ef8f6369e3fbc679689a55f3f79" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "4d9f67c6a363c98ed0bdaefac75bcf6f" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "4de782024372177c9dd854d532a2dd47" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "d91fbadbb5a5c96c07a0f959f2b07373" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 31662080, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28190720 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28200960 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 28211200 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 29358080 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 31651840 + } + ], + "md5sum": "6bd7934931c6b8d615e6baea96312477" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "fef04743a3b0530f5958373cdfb30bf3" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "9573ba181cdde60f49d3c1e2d813d04e" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "38aca88bbbd01f27c85cdb682dfc0c04" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "ee653fc119365a9c63edb6e54cf3a443" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "2fc8dfb72603beb221c404b79b151626" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "d51cc86a3f76618471b6c98af79a65b7" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "ec57fba00bd7d98d866c15f7f34a6182" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "46a12ddb2900649e6eb33a6136f77011" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "730bc23ef5b7cc99349d42a3167ab1d9" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 31662080, + "records": [ + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28190720 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28200960 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 28211200 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 29358080 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 31651840 + } + ], + "md5sum": "82ba52ba87e4320215707aa7a86c80c0" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "77a60584cfa89b5d3ba48132de9ce837" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "9a33db34a7ad452f882c2f5e55f6762c" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "5a34c4fce37d2e5ca59641bd17503fe6" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "3455b8e9056257c6a356a69652f2efd3" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "25ad89f503d2d1b2c2f3a12ad5867eea" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "fae1f6449163c1ede72af3fc0ac8255b" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "d8d9a145f6c1a769565b428da427d5f8" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "f86258d6572b00afe491d1bbff18d151" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "a09085e1e2a175d41c5dc6ce6bb6dfd9" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "4a37d509b9fac2560341861f909870b2" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "9331f59e435714e54c2dddd905425c14" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "808e4afc5002eefa46e9faec56c86ee9" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "8e192951b6c3432bb1e5068be0c85318" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "af17b5e71b2b469fb9572ef066fcf0c1" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "46a66dc2a927967c221cc9895e7a2309" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "ede0204e8f2b8dcae7f998d24a6102df" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "307cdab000cbca1e1d845046ce71ab8d" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "c56e433cdefcbb16dd31d6503314cbe9" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "e75d7b72f4c5a31fc6833dbed73a778b" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c71ea545a8ce0105fb2698af02f2efc9" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "042b7c6df2a21bf6eff5448eec4ed8fa" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "8a287cad6a18783ade4561f53d0fc733" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 29327360, + "records": [ + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 27033600 + } + ], + "md5sum": "a55528666f29ac19aae3401c49f63dc9" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "f880ee8874c116ede9ff0d48ee80aa8e" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "8054296c804f24b7e2a62d8b3438371f" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "c2579c67a8758122858ce1764639361b" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 31662080, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28190720 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 28200960 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 28211200 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 29358080 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 31651840 + } + ], + "md5sum": "1087afb7f68329299843b6e22ebc01b8" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "bf6ca26c2fe5b069196d21ac673738ef" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "cec0cbd7a955ccb43fa6c03c9eaf59a9" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "05e799c120363aa80df695769f4d3be8" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "cb992ae0f6fe87f36a64fd594cf701bf" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "30d748a9f00014704ee879596f39e328" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "e93dbcba1f3c33d5d1162405050c573c" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "99f500eb3b037d94d0b7e2dc3984dc2e" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "d2e2f40bc68746b475be279a1134ea02" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "bcd00a67c1dd9509f8ad310be31aea9a" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "35253e092053de8280979fb6362b530b" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "3eaad702e42fba3019624299a8e08839" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.28.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "a2a5e1380be797f19ea617bd891a63b6" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "5aa8272167b10f5355588f13bc874460" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "0ce8f30b1bc1bf0f951c1277b1b78ca0" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.29.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "8c57c06d92dddd1bc4c59c7990eceb44" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "ba634d8d7dc4ad74957388ce08a1c4ee" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "537005b903d7b9d5b9c572adb5cf17b8" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.30.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "42fc9a73f00aa6edf92803b384bfd21b" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 36700160, + "records": [ + { + "name": "model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 1792, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 36700160, + "byteOffset": 0 + } + ], + "md5sum": "47fa9189dc9fa38fa54dfcfbae069e69" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 73400320, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 640, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 73400320, + "byteOffset": 0 + } + ], + "md5sum": "ed7624941fe3fdce22b9f7fec12bca93" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 30494720, + "records": [ + { + "name": "model.layers.31.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 27033600 + }, + { + "name": "model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 112, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1146880, + "byteOffset": 27043840 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 40, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2293760, + "byteOffset": 28190720 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 10240, + "byteOffset": 30484480 + } + ], + "md5sum": "b915efc934adf8fda2a9e847973c2a2e" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 27033600, + "records": [ + { + "name": "model.layers.32.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.32.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + } + ], + "md5sum": "132877788a82c836c7285b8dce1c0804" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 27033600, + "records": [ + { + "name": "model.layers.33.self_attn.qkv_proj.q_weight", + "shape": [ + 640, + 6144 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 15728640, + "byteOffset": 0 + }, + { + "name": "model.layers.33.self_attn.qkv_proj.q_scale", + "shape": [ + 40, + 6144 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 491520, + "byteOffset": 15728640 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 512, + 5120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 10485760, + "byteOffset": 16220160 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 32, + 5120 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 327680, + "byteOffset": 26705920 + } + ], + "md5sum": "b7aa356aed3c305bed9a4b994979f237" + } + ] +} \ No newline at end of file