diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,6425 @@ +{ + "metadata": { + "ParamSize": 507, + "ParamBytes": 4766153728.0, + "BitsPerParam": 4.068522985105678 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 458752000, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 256000, + 448 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 458752000, + "byteOffset": 0 + } + ], + "md5sum": "85ddc730c8a7f5f548d3e5857de426bc" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "00cf549b61c1818f57c5da869563eff9" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "5a6b42787fe8cceb9747edffc2cec6a3" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 31911936, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 256000, + 28 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 14336000, + "byteOffset": 0 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 14336000 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 14343168 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 15145984 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 16751616 + }, + { + "name": "model.layers.0.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 16758784 + }, + { + "name": "model.layers.0.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 16765952 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 16773120 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 31453184 + } + ], + "md5sum": "7f0bc91f877db774d4cafe2f064d73cf" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 33266688, + "records": [ + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 0 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 7340032 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 7569408 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 7576576 + } + ], + "md5sum": "2dfd2bcb95e0e89bf975628108465d29" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "7b9c73b7945927ac0a375c8229da52ad" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "67a8610cdbc9077e829cab0f10b167a6" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "740225e4ce9ce4d779d6d2f3b0e2ea40" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 27575296, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 802816 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2408448 + }, + { + "name": "model.layers.1.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2415616 + }, + { + "name": "model.layers.1.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2422784 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 2429952 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 17110016 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 17568768 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 24908800 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 25145344 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 25948160 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27553792 + }, + { + "name": "model.layers.2.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27560960 + }, + { + "name": "model.layers.2.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27568128 + } + ], + "md5sum": "a749e2c09149d8b3f0b1a8663b7e17cd" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "22a4e019a6f0d56a494242bb00085969" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "f18c1ed4244a269bb187d0fe6c5d1411" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.3.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.3.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "192150e174eb20db725bd04eb85f4b6e" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "3d9fa73c791524180aa7a1e71ea9b0ed" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "25051002a3d2eb08b49fea307a3db9f6" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.4.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.4.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "352e5c58e68e9e5d806ce77178fc0674" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "ee25040a305fb4f67dd8656388290779" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "4abe092da1fecce188dd6669192ec8cc" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.5.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.5.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "b704819d8ab9ef6cac7c5a936bf0870e" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d30fcec49d243cb163439d6ed2e5a888" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "4d6c7153ff08f2ae4ab79f331481cbab" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.6.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.6.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "74347901b831a39d74d917b4d3cf3643" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "88a48a340d8102bb5a61ebbc97c31c6a" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 24313856, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 22708224 + } + ], + "md5sum": "6ada8676816106d6786323d93e798857" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "252bea5480011d17cc00192ad847f9a8" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "a1c510c0a0547085b682bbe3c49ed175" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.10.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.10.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "58df2a73346145abe870e46d7d782cc1" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "89d467a609151e2afbe6c7fb0f1f75ea" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "54653e042c5daa40e17b8c961b5d18d0" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.11.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.11.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "d817777c1e8e5f0a5e80b4bbafd9aeb8" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "659f91e7ad93b8c7f70cf11c7a255f4e" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "519fe6793be6039dddc8736a975c123d" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.12.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.12.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "77d7b6204515535e0293eb6765cd064a" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "8d12bf225ae00fe6b39d99f0f81c633b" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "3fb2f7254ab06c61ca3e1899e05db753" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.13.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.13.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "0f308864db5827a294ae1694a422f861" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d189726509ccb3049c17e6bfc8d34dc8" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "759a1d4a26130b78dfe3cc369819066b" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.14.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.14.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "a14729ba0e6a991a57c1c9b93f6f4fef" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "89d06177461be77c3488639e0dae4a1b" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "84124abc59ca33f87c9c315dfdd8320a" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.15.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.15.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "01e5f1f36f7d0930efcbb825ba44d8be" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "53489b9553f5eb8d802b3f0f3046d8c0" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "d778563f031247d4d7227f5d74119569" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.16.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.16.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "630f3d721c171b0a26ed6a3591ba69b2" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "5462cbc19420547a2e6b54f0190163ff" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "c2e9d8488708770e04bd94d5a9d82a33" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.17.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.17.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "aaeb47576326ac0bf2eb6f9a7eaf934a" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "87cc6b6ec3f8d2d7677b30f0280c0e2c" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "243154c8fac8e9110feabbf80199a322" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.18.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.18.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "b2efbee647d1954e05a7dbb984ce87b1" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d91cfe945ceb15146455e8184d68d6e6" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "ce3ecc8672c5e39aa31fe4e326828901" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.19.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.19.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "1585c624b1b6fe4c6afbed42c1eb2d4b" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 22708224, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + } + ], + "md5sum": "38cf6f88ea787bdd3f91e89bc8be7c84" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "ae4bb8b8a7695d061d943cc125a2851e" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "30e4d679ee723d6bacf1a8bf80fc5f87" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "42fb8f55b79b54bf34cc95d7a6dbda8c" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 33546240, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 15138816 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 15145984 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 15948800 + }, + { + "name": "model.layers.7.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 15955968 + }, + { + "name": "model.layers.7.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 15963136 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 15970304 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 15977472 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 16780288 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 18385920 + }, + { + "name": "model.layers.8.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 18393088 + }, + { + "name": "model.layers.8.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 18400256 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 18407424 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 33087488 + } + ], + "md5sum": "0e7a25eecbf0a16fa6689e814d410c89" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 33266688, + "records": [ + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 0 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 7340032 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 7569408 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 7576576 + } + ], + "md5sum": "cc52fa7d8e932e6bee45c0c7f0b2261c" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "d65ecda28c7b24f1b745d9c93d40f651" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "5c161d2e18bc9f622d8cf537c5bde66a" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "26bb4aeaee24f0b71665b852c969e977" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 27575296, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 802816 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2408448 + }, + { + "name": "model.layers.9.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2415616 + }, + { + "name": "model.layers.9.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2422784 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 2429952 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 17110016 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 17568768 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 24908800 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 25145344 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 25948160 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27553792 + }, + { + "name": "model.layers.20.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27560960 + }, + { + "name": "model.layers.20.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27568128 + } + ], + "md5sum": "94cf7d0649a61224cf200241da97244d" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 33266688, + "records": [ + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 7340032 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 7569408 + }, + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 7576576 + } + ], + "md5sum": "a2fbd7e52d56417acbff44487adbe471" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "b977640bef40db16ad1c84ebe17b5db7" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "38a11314fb661334854f52db656cf840" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "994b44ed5b1d4a821270961d52fe7f50" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 27575296, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 0 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 802816 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2408448 + }, + { + "name": "model.layers.21.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2415616 + }, + { + "name": "model.layers.21.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 2422784 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 2429952 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 17110016 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 17568768 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 24908800 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 25145344 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 25948160 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27553792 + }, + { + "name": "model.layers.22.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27560960 + }, + { + "name": "model.layers.22.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 27568128 + } + ], + "md5sum": "d7ccb407731355f65d0d12695ad945ec" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "2ab13ccb329fe5d8d7adcf76cdf6651f" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "176537d806f5012f5f2f69f6caa17c3a" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.23.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.23.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "3c1a33c19abb6ce4793f5762365a8268" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "abc84d06a0e9ca1df8a3bd90137ced12" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "e33906afaa8e93d4d266a476bf527613" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.24.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.24.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "fd236653e334caaa5ac2869c2143f596" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "eeecdf12f947b2fc08a66eca7b12b5b6" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "6981fd2af8b4ce8ee8be92fdc1d46136" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.25.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.25.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "e16c5824c9f2b85abd83b6817315d2c2" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "e00aaf716b3ddd7f9f837fd83822caaa" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "518a4983e7c1f0da64acbab345902928" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.26.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.26.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "3a2dde70f6a32434a096980643755b08" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "ac84c6d006ae3b01be8555c35e5cc5f0" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "799a6f4686a01746670a76ee142c22a1" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.27.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.27.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "70e897415fb1af9b52b166a883267c08" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "c215174121b5033f3d9d8ae0c673b0b5" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "e200e77a4f24adb199a86ce1ddccca57" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.28.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.28.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "ac59b121fccdba84d17187ddd32dfac5" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d4dd7a11bf98c9d4a7949b6ceda9526d" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "36085b823838fa32444466855c45e0ee" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.28.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.29.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.29.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "e9b60154b3993ccf410cb9f296dd23c4" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "1dbd1de258659ab0d7c03d2995127b13" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "97957e8b10f5c00ee82307b0d63022ce" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.29.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.30.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.30.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "8fed37e94f59ffda8177b898bebc8844" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "82ac10ed46a69f6f5a897aec65b3cce5" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "a187852a1fa995501041923b0ec3e309" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.30.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.31.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.31.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "a535a4db95581ec3b05bf7765747e7be" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.32.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "0b67c3c3da3e7aff707bb0025d47541b" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 24313856, + "records": [ + { + "name": "model.layers.31.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.32.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 22708224 + } + ], + "md5sum": "3d9122887d019cecad6c80ab4349a41a" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.32.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d169e96c6657a21ff66f583699b0d808" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.33.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "978a8c31d2afa23925dbccdc39f17151" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.33.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "8f17dc58583e2de39873221e43fa4085" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 25976832, + "records": [ + { + "name": "model.layers.32.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.32.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.32.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.32.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.32.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.32.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23518208 + }, + { + "name": "model.layers.32.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23525376 + }, + { + "name": "model.layers.32.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23532544 + }, + { + "name": "model.layers.33.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 23539712 + }, + { + "name": "model.layers.33.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 23546880 + }, + { + "name": "model.layers.33.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 24349696 + }, + { + "name": "model.layers.33.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25955328 + }, + { + "name": "model.layers.33.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25962496 + }, + { + "name": "model.layers.33.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25969664 + } + ], + "md5sum": "53dc0ec7db6bb87a6dab3d6cf37dec8f" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.34.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "fd404359d91c6e2628d5d81ff72f1d72" + }, + { + "dataPath": "params_shard_102.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.34.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "c0831e2b894e36a9d94bc53b3c1c54e5" + }, + { + "dataPath": "params_shard_103.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.33.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.33.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.33.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.34.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.34.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.34.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.34.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.34.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.34.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "a5c16a6bba9e707e82a9b41cbca5e8a7" + }, + { + "dataPath": "params_shard_104.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.35.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "837f40900b1a6981322daf53f301c18f" + }, + { + "dataPath": "params_shard_105.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.35.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "95072e4e6bcc56d489b668cc1325f0e1" + }, + { + "dataPath": "params_shard_106.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.34.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.34.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.34.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.35.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.35.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.35.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.35.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.35.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.35.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "2c2455bffc1c5df5bf708f1d45b8f9f4" + }, + { + "dataPath": "params_shard_107.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.36.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "713007a1abc77367e1e77c6e97fc2536" + }, + { + "dataPath": "params_shard_108.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.36.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "ad92dce884724fe904d185e28023b130" + }, + { + "dataPath": "params_shard_109.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.35.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.35.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.35.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.36.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.36.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.36.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.36.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.36.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.36.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "22c179c2f9fef29359f51eda8e1c1734" + }, + { + "dataPath": "params_shard_110.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.37.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d28b236cdd1eac79e15fff3edad7fa3d" + }, + { + "dataPath": "params_shard_111.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.37.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "853ab05423b3618c391e984201bdf04c" + }, + { + "dataPath": "params_shard_112.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.36.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.36.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.36.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.37.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.37.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.37.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.37.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.37.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.37.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "e16e829a3c5aea1929de40d1ddf081b4" + }, + { + "dataPath": "params_shard_113.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.38.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "ad5719c33b79273152a5ab80d056a196" + }, + { + "dataPath": "params_shard_114.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.38.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "4a9c21d34321eb7b32850573a2dbc4f6" + }, + { + "dataPath": "params_shard_115.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.37.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.37.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.37.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.38.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.38.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.38.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.38.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.38.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.38.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "3336a553cd8ea312581a5b6b294e1cf2" + }, + { + "dataPath": "params_shard_116.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.39.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "2040b5b801732a019733d4c603a68df7" + }, + { + "dataPath": "params_shard_117.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.39.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "14d1b3b6a04d773b5fadcc904400a0d8" + }, + { + "dataPath": "params_shard_118.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.38.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.38.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.38.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.39.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.39.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.39.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.39.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.39.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.39.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "4aff3d4c4e44f852029fa708f04c8133" + }, + { + "dataPath": "params_shard_119.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.40.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "00f0ccd0a30e90f0c12afc50dd894851" + }, + { + "dataPath": "params_shard_120.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.40.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "33f1d394579d76fef312bec84dbd8f15" + }, + { + "dataPath": "params_shard_121.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.39.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.39.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.39.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.40.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.40.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.40.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.40.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.40.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.40.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "17134e26594d6d357717ed463c3d614b" + }, + { + "dataPath": "params_shard_122.bin", + "format": "raw-shard", + "nbytes": 25690112, + "records": [ + { + "name": "model.layers.41.mlp.down_proj.q_weight", + "shape": [ + 1792, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 25690112, + "byteOffset": 0 + } + ], + "md5sum": "d572bcebaf4ba71c337ad3b29a9175f5" + }, + { + "dataPath": "params_shard_123.bin", + "format": "raw-shard", + "nbytes": 51380224, + "records": [ + { + "name": "model.layers.41.mlp.gate_up_proj.q_weight", + "shape": [ + 448, + 28672 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 51380224, + "byteOffset": 0 + } + ], + "md5sum": "beacc186df3409bf63892704b35bed5c" + }, + { + "dataPath": "params_shard_124.bin", + "format": "raw-shard", + "nbytes": 25145344, + "records": [ + { + "name": "model.layers.40.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.40.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.40.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.40.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.layers.41.input_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + }, + { + "name": "model.layers.41.mlp.down_proj.q_scale", + "shape": [ + 112, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 802816, + "byteOffset": 22715392 + }, + { + "name": "model.layers.41.mlp.gate_up_proj.q_scale", + "shape": [ + 28, + 28672 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1605632, + "byteOffset": 23518208 + }, + { + "name": "model.layers.41.post_attention_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25123840 + }, + { + "name": "model.layers.41.post_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25131008 + }, + { + "name": "model.layers.41.pre_feedforward_layernorm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 25138176 + } + ], + "md5sum": "1a5603b7c759b444c7ed3d87a32f53a4" + }, + { + "dataPath": "params_shard_125.bin", + "format": "raw-shard", + "nbytes": 22715392, + "records": [ + { + "name": "model.layers.41.self_attn.qkv_proj.q_weight", + "shape": [ + 448, + 8192 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 14680064, + "byteOffset": 0 + }, + { + "name": "model.layers.41.self_attn.qkv_proj.q_scale", + "shape": [ + 28, + 8192 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 458752, + "byteOffset": 14680064 + }, + { + "name": "model.layers.41.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3584 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 7340032, + "byteOffset": 15138816 + }, + { + "name": "model.layers.41.self_attn.o_proj.q_scale", + "shape": [ + 32, + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 229376, + "byteOffset": 22478848 + }, + { + "name": "model.norm.weight", + "shape": [ + 3584 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 7168, + "byteOffset": 22708224 + } + ], + "md5sum": "e5b3d461a88f0503aaba10b62fa92004" + } + ] +} \ No newline at end of file