diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3553 @@ +{ + "metadata": { + "ParamSize": 323, + "ParamBytes": 203614080.0, + "BitsPerParam": 4.501983300477319 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 33459840, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 49152, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 23592960, + "byteOffset": 0 + }, + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 49152, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2949120, + "byteOffset": 23592960 + }, + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 26542080 + }, + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 26544000 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 27772800 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 27926400 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 30384000 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 30691200 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 30693120 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 31461120 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 31557120 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 32017920 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 32075520 + }, + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 32077440 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 33306240 + } + ], + "md5sum": "9ef2c61216184d170dfcd73484f8c805" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 0 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 2457600 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 2764800 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 2766720 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 3534720 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 3630720 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 4091520 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 4149120 + }, + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 4151040 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 5379840 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 5533440 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 7991040 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 8298240 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 8300160 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 9068160 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 9164160 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 9624960 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 9682560 + }, + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 9684480 + }, + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 10913280 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 11066880 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 13524480 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 13831680 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 13833600 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 14601600 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 14697600 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 15158400 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 15216000 + }, + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 15217920 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 16446720 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 16600320 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 19057920 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 19365120 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 19367040 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 20135040 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 20231040 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 20691840 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 20749440 + }, + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 20751360 + }, + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 21980160 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 22133760 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 24591360 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 24898560 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 24900480 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 25668480 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 25764480 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 26225280 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 26282880 + }, + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 26284800 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 27513600 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 27667200 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 30124800 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 30432000 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 30433920 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 31201920 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 31297920 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 31758720 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 31816320 + }, + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 31818240 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 33047040 + } + ], + "md5sum": "de45b4820816c3296d57c2dee9742a4c" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 0 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 2457600 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 2764800 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 2766720 + }, + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 3534720 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 3630720 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 4091520 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 4149120 + }, + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 4151040 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 5379840 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 5533440 + }, + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 7991040 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 8298240 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 8300160 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 9068160 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 9164160 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 9624960 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 9682560 + }, + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 9684480 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 10913280 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 11066880 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 13524480 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 13831680 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 13833600 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 14601600 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 14697600 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 15158400 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 15216000 + }, + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 15217920 + }, + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 16446720 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 16600320 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 19057920 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 19365120 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 19367040 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 20135040 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 20231040 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 20691840 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 20749440 + }, + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 20751360 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 21980160 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 22133760 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 24591360 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 24898560 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 24900480 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 25668480 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 25764480 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 26225280 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 26282880 + }, + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 26284800 + }, + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 27513600 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 27667200 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 30124800 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 30432000 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 30433920 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 31201920 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 31297920 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 31758720 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 31816320 + }, + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 31818240 + }, + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 33047040 + } + ], + "md5sum": "ba7137d605ce572f90c81db091a36e90" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 0 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 2457600 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 2764800 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 2766720 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 3534720 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 3630720 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 4091520 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 4149120 + }, + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 4151040 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 5379840 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 5533440 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 7991040 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 8298240 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 8300160 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 9068160 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 9164160 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 9624960 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 9682560 + }, + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 9684480 + }, + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 10913280 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 11066880 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 13524480 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 13831680 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 13833600 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 14601600 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 14697600 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 15158400 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 15216000 + }, + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 15217920 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 16446720 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 16600320 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 19057920 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 19365120 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 19367040 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 20135040 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 20231040 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 20691840 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 20749440 + }, + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 20751360 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 21980160 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 22133760 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 24591360 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 24898560 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 24900480 + }, + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 25668480 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 25764480 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 26225280 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 26282880 + }, + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 26284800 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 27513600 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 27667200 + }, + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 30124800 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 30432000 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 30433920 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 31201920 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 31297920 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 31758720 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 31816320 + }, + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 31818240 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 33047040 + } + ], + "md5sum": "cc596189cecaf6498b744e6029c98d02" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 0 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 2457600 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 2764800 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 2766720 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 3534720 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 3630720 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 4091520 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 4149120 + }, + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 4151040 + }, + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 5379840 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 5533440 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 7991040 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 8298240 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 8300160 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 9068160 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 9164160 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 9624960 + }, + { + "name": "model.layers.28.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 9682560 + }, + { + "name": "model.layers.28.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 9684480 + }, + { + "name": "model.layers.28.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 10913280 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 11066880 + }, + { + "name": "model.layers.28.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 13524480 + }, + { + "name": "model.layers.28.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 13831680 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 13833600 + }, + { + "name": "model.layers.28.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 14601600 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 14697600 + }, + { + "name": "model.layers.28.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 15158400 + }, + { + "name": "model.layers.29.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 15216000 + }, + { + "name": "model.layers.29.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 15217920 + }, + { + "name": "model.layers.29.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 16446720 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 16600320 + }, + { + "name": "model.layers.29.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 19057920 + }, + { + "name": "model.layers.29.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 19365120 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 19367040 + }, + { + "name": "model.layers.29.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 20135040 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 20231040 + }, + { + "name": "model.layers.29.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 20691840 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 20749440 + }, + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 20751360 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 21980160 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 22133760 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 24591360 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 24898560 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 24900480 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 25668480 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 25764480 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 26225280 + }, + { + "name": "model.layers.30.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 26282880 + }, + { + "name": "model.layers.30.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 26284800 + }, + { + "name": "model.layers.30.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 27513600 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 27667200 + }, + { + "name": "model.layers.30.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 30124800 + }, + { + "name": "model.layers.30.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 30432000 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 30433920 + }, + { + "name": "model.layers.30.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 31201920 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 31297920 + }, + { + "name": "model.layers.30.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 31758720 + }, + { + "name": "model.layers.31.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 31816320 + }, + { + "name": "model.layers.31.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 31818240 + }, + { + "name": "model.layers.31.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 33047040 + } + ], + "md5sum": "a173bd5accb573f77d82558f0dcec89c" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 33200640, + "records": [ + { + "name": "model.layers.31.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 0 + }, + { + "name": "model.layers.31.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 2457600 + }, + { + "name": "model.layers.31.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 2764800 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 2766720 + }, + { + "name": "model.layers.31.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 3534720 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 3630720 + }, + { + "name": "model.layers.31.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 4091520 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 4149120 + }, + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 4151040 + }, + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 5379840 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 5533440 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 7991040 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 8298240 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 8300160 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 9068160 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 9164160 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 9624960 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 9682560 + }, + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 9684480 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 10913280 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 11066880 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 13524480 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 13831680 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 13833600 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 14601600 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 14697600 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 15158400 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 15216000 + }, + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 15217920 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 16446720 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 16600320 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 19057920 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 19365120 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 19367040 + }, + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 20135040 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 20231040 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 20691840 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 20749440 + }, + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 20751360 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 21980160 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 22133760 + }, + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 24591360 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 24898560 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 24900480 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 25668480 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 25764480 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 26225280 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 26282880 + }, + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 26284800 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 27513600 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 27667200 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 30124800 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 30432000 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 30433920 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 31201920 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 31297920 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 31758720 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 31816320 + }, + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 960, + 320 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 1228800, + "byteOffset": 31818240 + }, + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 960, + 80 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 153600, + "byteOffset": 33047040 + } + ], + "md5sum": "87445e0fb92453f8d4759db66a819700" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 4151040, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 5120, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 2457600, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 5120, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 307200, + "byteOffset": 2457600 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 2764800 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 1600, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 768000, + "byteOffset": 2766720 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 1600, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 96000, + "byteOffset": 3534720 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 960, + 120 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 460800, + "byteOffset": 3630720 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 960, + 30 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 57600, + "byteOffset": 4091520 + }, + { + "name": "model.norm.weight", + "shape": [ + 960 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 1920, + "byteOffset": 4149120 + } + ], + "md5sum": "6f5b8a08433da9b216cc8047776a81f4" + } + ] +} \ No newline at end of file