diff --git "a/ndarray-cache.json" "b/ndarray-cache.json" new file mode 100644--- /dev/null +++ "b/ndarray-cache.json" @@ -0,0 +1,3881 @@ +{ + "metadata": { + "ParamSize": 283, + "ParamBytes": 4802697216.0, + "BitsPerParam": 4.500235859834132 + }, + "records": [ + { + "dataPath": "params_shard_0.bin", + "format": "raw-shard", + "nbytes": 393216000, + "records": [ + { + "name": "model.embed_tokens.q_weight", + "shape": [ + 256000, + 384 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 393216000, + "byteOffset": 0 + } + ], + "md5sum": "752374fe75aeec625f154bdba2af2447" + }, + { + "dataPath": "params_shard_1.bin", + "format": "raw-shard", + "nbytes": 49152000, + "records": [ + { + "name": "model.embed_tokens.q_scale", + "shape": [ + 256000, + 96 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 49152000, + "byteOffset": 0 + } + ], + "md5sum": "67ce689d7e81e4d7e847a5fb1dd51983" + }, + { + "dataPath": "params_shard_2.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.0.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "746ad94dbbc5e5c3946d13b5bf727e90" + }, + { + "dataPath": "params_shard_3.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.0.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "49adfd54eb8b3ab4e1e89e9dd3cdf47e" + }, + { + "dataPath": "params_shard_4.bin", + "format": "raw-shard", + "nbytes": 33042432, + "records": [ + { + "name": "model.layers.0.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 0 + }, + { + "name": "model.layers.0.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 6144 + }, + { + "name": "model.layers.0.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4724736 + }, + { + "name": "model.layers.0.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14161920 + }, + { + "name": "model.layers.0.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14168064 + } + ], + "md5sum": "275395bf4fb959d0a8c89cfb24762dab" + }, + { + "dataPath": "params_shard_5.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.1.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "a1c5a7d8333657710b4dcaf2f8554be2" + }, + { + "dataPath": "params_shard_6.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.1.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "f8a5be8a43625f7dae1d6a320dc985a8" + }, + { + "dataPath": "params_shard_7.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.1.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "472d2a7b0930900e3a25c6e78988b159" + }, + { + "dataPath": "params_shard_8.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "17362291fc07e4cadc28d9443c254f84" + }, + { + "dataPath": "params_shard_9.bin", + "format": "raw-shard", + "nbytes": 33048576, + "records": [ + { + "name": "model.layers.0.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.0.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.1.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.1.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.1.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.1.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.1.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.1.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + }, + { + "name": "model.layers.2.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 33042432 + } + ], + "md5sum": "20c05b2a3fba919923aaf12970efa0ea" + }, + { + "dataPath": "params_shard_10.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.2.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "deb3af66351b2089ae9f8696f3519d6b" + }, + { + "dataPath": "params_shard_11.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.2.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.2.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.2.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.2.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "7801a6d91f0b4f1931d8a1ca8339ad2b" + }, + { + "dataPath": "params_shard_12.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.3.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "b73413de2a24bdf9a702b8be703a3915" + }, + { + "dataPath": "params_shard_13.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.3.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "9233aeedb818c823fdbc8d63ea04e413" + }, + { + "dataPath": "params_shard_14.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.3.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "c6207691bcd18659d1b0e384c061120d" + }, + { + "dataPath": "params_shard_15.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "d6fabb1b11e1eb51a297ffa3b684116a" + }, + { + "dataPath": "params_shard_16.bin", + "format": "raw-shard", + "nbytes": 33048576, + "records": [ + { + "name": "model.layers.2.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.2.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.3.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.3.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.3.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.3.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.3.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.3.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + }, + { + "name": "model.layers.4.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 33042432 + } + ], + "md5sum": "9b7b81f3b8a8ce23e8362ca840e16727" + }, + { + "dataPath": "params_shard_17.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.4.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "ee9fc39bfbf7689188bfa0921485eb3a" + }, + { + "dataPath": "params_shard_18.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.4.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.4.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.4.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.4.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "d723cc19fe2c7583878c2892548dfaf8" + }, + { + "dataPath": "params_shard_19.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.5.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "4a8043e0455315b8f52fff4e9e2d4d94" + }, + { + "dataPath": "params_shard_20.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.5.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "374d4dc68dfdf03c8240a7342d0df8a2" + }, + { + "dataPath": "params_shard_21.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.5.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "6aacf0adf550fb907fc87fb9a948cf18" + }, + { + "dataPath": "params_shard_22.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "9a249cce2324da725e92f0a97b7f0236" + }, + { + "dataPath": "params_shard_23.bin", + "format": "raw-shard", + "nbytes": 33042432, + "records": [ + { + "name": "model.layers.4.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.4.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.5.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.5.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.5.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.5.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.5.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.5.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + } + ], + "md5sum": "289977dbbcf0627a243efdee02e33637" + }, + { + "dataPath": "params_shard_24.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.10.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "0e70c5559843d0f3f6e4aeb70f8cb16c" + }, + { + "dataPath": "params_shard_25.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.10.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "de718bbb230ac8e592a71b981abd5547" + }, + { + "dataPath": "params_shard_26.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.10.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "17850eaea95724f6f3c09c7dcd974454" + }, + { + "dataPath": "params_shard_27.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "b3cffd71d62beafd2e3ac066fcf052aa" + }, + { + "dataPath": "params_shard_28.bin", + "format": "raw-shard", + "nbytes": 33048576, + "records": [ + { + "name": "model.layers.6.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.6.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.10.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.10.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.10.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.10.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.10.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.10.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + }, + { + "name": "model.layers.11.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 33042432 + } + ], + "md5sum": "f7c85ef13a43237fa5943ed990b13616" + }, + { + "dataPath": "params_shard_29.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.11.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "a6fd89d0f77762710e7dfc0ea3f865a1" + }, + { + "dataPath": "params_shard_30.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.11.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.11.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.11.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.11.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "18093e8c9466adbd15fd8f76296eb0da" + }, + { + "dataPath": "params_shard_31.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.12.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "b063b6cfc160691f5584a9cf41739a0d" + }, + { + "dataPath": "params_shard_32.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.12.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "6d4e2c09057135b2dd07b60bccd375b6" + }, + { + "dataPath": "params_shard_33.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.12.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "3034411e64e200cff10cd54a935986ac" + }, + { + "dataPath": "params_shard_34.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "f5f276e4f4c0142e194de0bff2286947" + }, + { + "dataPath": "params_shard_35.bin", + "format": "raw-shard", + "nbytes": 33048576, + "records": [ + { + "name": "model.layers.11.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.11.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.12.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.12.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.12.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.12.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.12.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.12.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + }, + { + "name": "model.layers.13.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 33042432 + } + ], + "md5sum": "659bb12be1ee9cc3754feddb1357773a" + }, + { + "dataPath": "params_shard_36.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.13.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "d03356ef88c3d69bd33a5fa1ca2c6e19" + }, + { + "dataPath": "params_shard_37.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.13.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.13.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.13.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.13.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "238920c9d8c5d46753599401e4a4979b" + }, + { + "dataPath": "params_shard_38.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.14.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "d0a82389fc83c2991b349df03e29713e" + }, + { + "dataPath": "params_shard_39.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.14.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "ea5d93cbc3fae66165fe0f13224abaaf" + }, + { + "dataPath": "params_shard_40.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.14.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "1f81a9b75a3e7e237668c90d852e4ec3" + }, + { + "dataPath": "params_shard_41.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "50bb1467a0ddba10596ea56fa3abdc7f" + }, + { + "dataPath": "params_shard_42.bin", + "format": "raw-shard", + "nbytes": 33042432, + "records": [ + { + "name": "model.layers.13.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.13.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.14.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.14.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.14.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.14.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.14.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.14.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + } + ], + "md5sum": "e761bb62069ee6aeca9d6887ea1d8857" + }, + { + "dataPath": "params_shard_43.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.6.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "4a05c49ebed9b7d718073b183e8089a7" + }, + { + "dataPath": "params_shard_44.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.6.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "aeb1493a8fbe560c6837e0f5e691f668" + }, + { + "dataPath": "params_shard_45.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.7.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "95fa6545ae01c6ba07721595a1cf2a29" + }, + { + "dataPath": "params_shard_46.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "683f161e289529cb7201f33ca58ffcea" + }, + { + "dataPath": "params_shard_47.bin", + "format": "raw-shard", + "nbytes": 28329984, + "records": [ + { + "name": "model.layers.15.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.15.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.6.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.6.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.6.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.6.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.7.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.7.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 23611392 + } + ], + "md5sum": "da47319eaa593ffd13886df802e1da66" + }, + { + "dataPath": "params_shard_48.bin", + "format": "raw-shard", + "nbytes": 30676992, + "records": [ + { + "name": "model.layers.7.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 0 + }, + { + "name": "model.layers.7.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 9443328 + }, + { + "name": "model.layers.7.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 28317696 + } + ], + "md5sum": "402604a556d211e06d2104fe58ff96c3" + }, + { + "dataPath": "params_shard_49.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.8.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "61536dad74dbab013377a94eeb70a14d" + }, + { + "dataPath": "params_shard_50.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.8.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "9b597d135ccf9004b6fbd04764ce4c70" + }, + { + "dataPath": "params_shard_51.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.8.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "232c9ae1cad87524740d1ba388f3039d" + }, + { + "dataPath": "params_shard_52.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "62134d1562e43cdce27ae77c264f1e81" + }, + { + "dataPath": "params_shard_53.bin", + "format": "raw-shard", + "nbytes": 30689280, + "records": [ + { + "name": "model.layers.7.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.7.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.8.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 7077888 + }, + { + "name": "model.layers.8.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 7084032 + }, + { + "name": "model.layers.8.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 11802624 + }, + { + "name": "model.layers.8.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 21239808 + }, + { + "name": "model.layers.8.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21245952 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 23605248 + }, + { + "name": "model.layers.8.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 29896704 + }, + { + "name": "model.layers.9.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 30683136 + } + ], + "md5sum": "88c532b62aa545e0953d8dc4cafbba5b" + }, + { + "dataPath": "params_shard_54.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.9.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "d300d64fe443737a8c98b072b8912cea" + }, + { + "dataPath": "params_shard_55.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.9.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.9.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.9.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.9.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "9cf938fc1573f90129b2412574b8d9e7" + }, + { + "dataPath": "params_shard_56.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.15.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "103abf44635b18b7bb456662175bb5aa" + }, + { + "dataPath": "params_shard_57.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.15.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "5e516279b5f807aeb382b85d12b79da6" + }, + { + "dataPath": "params_shard_58.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.16.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "a5b65d598ca50a2428d201280bae33eb" + }, + { + "dataPath": "params_shard_59.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "28dbcc483766ec6b0c426b53a87eb5e1" + }, + { + "dataPath": "params_shard_60.bin", + "format": "raw-shard", + "nbytes": 28329984, + "records": [ + { + "name": "model.layers.9.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.9.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.15.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.15.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.15.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.15.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.16.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.16.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 23611392 + } + ], + "md5sum": "2dc22bbece44fa639ad72db56c28acf9" + }, + { + "dataPath": "params_shard_61.bin", + "format": "raw-shard", + "nbytes": 30676992, + "records": [ + { + "name": "model.layers.16.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 0 + }, + { + "name": "model.layers.16.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 9443328 + }, + { + "name": "model.layers.16.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 28317696 + } + ], + "md5sum": "ee563289d28ecd5f03bdd3b452bc1432" + }, + { + "dataPath": "params_shard_62.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.17.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "44964973b881155e5f588ab19d677420" + }, + { + "dataPath": "params_shard_63.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.17.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "ee430d7c3ff86f5cd41188eac6322a52" + }, + { + "dataPath": "params_shard_64.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.17.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "e43bdc9900d919f8d48eca61049172ad" + }, + { + "dataPath": "params_shard_65.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "02828fe9d6d6bf4d930867c4d005a1c5" + }, + { + "dataPath": "params_shard_66.bin", + "format": "raw-shard", + "nbytes": 30689280, + "records": [ + { + "name": "model.layers.16.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.16.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.17.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 7077888 + }, + { + "name": "model.layers.17.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 7084032 + }, + { + "name": "model.layers.17.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 11802624 + }, + { + "name": "model.layers.17.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 21239808 + }, + { + "name": "model.layers.17.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21245952 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 23605248 + }, + { + "name": "model.layers.17.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 29896704 + }, + { + "name": "model.layers.18.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 30683136 + } + ], + "md5sum": "6e6ede6708786b529bc2bd386998d636" + }, + { + "dataPath": "params_shard_67.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.18.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "67d320c0608cedb1c0b6ddaca05529e4" + }, + { + "dataPath": "params_shard_68.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.18.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.18.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.18.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.18.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "954c2acf938fea71682295d2f5fb0268" + }, + { + "dataPath": "params_shard_69.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.19.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "272587536bcdaaced3b76153138012f2" + }, + { + "dataPath": "params_shard_70.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.19.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "8026f1ef2c15eaa109643418c1091608" + }, + { + "dataPath": "params_shard_71.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.19.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "ce35d8725f21746904c59c40b521560a" + }, + { + "dataPath": "params_shard_72.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "bc3655dbc318a34cc4b8d651e3e15b6f" + }, + { + "dataPath": "params_shard_73.bin", + "format": "raw-shard", + "nbytes": 33048576, + "records": [ + { + "name": "model.layers.18.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.18.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.19.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.19.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.19.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.19.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.19.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.19.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + }, + { + "name": "model.layers.20.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 33042432 + } + ], + "md5sum": "4bc7af89c43b230da07154be89392246" + }, + { + "dataPath": "params_shard_74.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.20.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "8332352971e7af2cab9e24bfdb1359ab" + }, + { + "dataPath": "params_shard_75.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.20.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.20.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.20.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.20.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "cddc31ae3164b327927d853519c99caf" + }, + { + "dataPath": "params_shard_76.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.21.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "84cbbb70c18bb2c8e85f870bc8d653e5" + }, + { + "dataPath": "params_shard_77.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.21.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "607b26798107e4b0fa7c3b1f49fda4f1" + }, + { + "dataPath": "params_shard_78.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.21.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "7003e513bf098f7b563fb2e17aa60cf5" + }, + { + "dataPath": "params_shard_79.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "08d04ed6d828c611bf911336e2ec3c24" + }, + { + "dataPath": "params_shard_80.bin", + "format": "raw-shard", + "nbytes": 33048576, + "records": [ + { + "name": "model.layers.20.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.20.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.21.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.21.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.21.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.21.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.21.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.21.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + }, + { + "name": "model.layers.22.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 33042432 + } + ], + "md5sum": "93e8116fc5a867e452624fd46feda2f6" + }, + { + "dataPath": "params_shard_81.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.22.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "960c6a79997443b10271b08cc0cedab4" + }, + { + "dataPath": "params_shard_82.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.22.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.22.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.22.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.22.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "8277c7c01db1e7f0ef83f36415f5c5c3" + }, + { + "dataPath": "params_shard_83.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.23.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "adf2acc291412d70bef1def2858d27ec" + }, + { + "dataPath": "params_shard_84.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.23.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "6dd2ca671c40b63024612c503f646646" + }, + { + "dataPath": "params_shard_85.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.23.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "8068db63028fdcbf58d9c115e74189df" + }, + { + "dataPath": "params_shard_86.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "444089513c22aeba4cb891d614382a15" + }, + { + "dataPath": "params_shard_87.bin", + "format": "raw-shard", + "nbytes": 33042432, + "records": [ + { + "name": "model.layers.22.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.22.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.23.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.23.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.23.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.23.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.23.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 23605248 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 25964544 + }, + { + "name": "model.layers.23.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 32256000 + } + ], + "md5sum": "0f940ab38043aed006d96d3188d0cae3" + }, + { + "dataPath": "params_shard_88.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.24.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "6f701fd42867ede64499f94735822adb" + }, + { + "dataPath": "params_shard_89.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.24.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "7b4b65f4fb128f3ac1d0f7d6e8229b87" + }, + { + "dataPath": "params_shard_90.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.25.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "118cc7806325ee0aca5c51e6ed610325" + }, + { + "dataPath": "params_shard_91.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "4b9d8697d5b03b77e8a5b2c717feb9cf" + }, + { + "dataPath": "params_shard_92.bin", + "format": "raw-shard", + "nbytes": 28329984, + "records": [ + { + "name": "model.layers.24.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.24.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.layers.24.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.24.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 9443328 + }, + { + "name": "model.layers.24.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 14161920 + }, + { + "name": "model.layers.24.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23599104 + }, + { + "name": "model.layers.25.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 23605248 + }, + { + "name": "model.layers.25.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 23611392 + } + ], + "md5sum": "4931e61db7c99f002a56815c1cc6710d" + }, + { + "dataPath": "params_shard_93.bin", + "format": "raw-shard", + "nbytes": 30676992, + "records": [ + { + "name": "model.layers.25.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 0 + }, + { + "name": "model.layers.25.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 9443328 + }, + { + "name": "model.layers.25.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 28317696 + } + ], + "md5sum": "5605e054ee014037eb1b1cba8eeca4bd" + }, + { + "dataPath": "params_shard_94.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.26.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "c81766fbfdb81748a71a02d6dddaae2c" + }, + { + "dataPath": "params_shard_95.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.26.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "7f341d507f778de05dcb0488b6c400a2" + }, + { + "dataPath": "params_shard_96.bin", + "format": "raw-shard", + "nbytes": 18874368, + "records": [ + { + "name": "model.layers.26.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 0 + } + ], + "md5sum": "ad7262dab759a60a8095f5e0fd741651" + }, + { + "dataPath": "params_shard_97.bin", + "format": "raw-shard", + "nbytes": 37748736, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_weight", + "shape": [ + 3072, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 37748736, + "byteOffset": 0 + } + ], + "md5sum": "3e16173cea5901db081dfb9ed64aa10b" + }, + { + "dataPath": "params_shard_98.bin", + "format": "raw-shard", + "nbytes": 30689280, + "records": [ + { + "name": "model.layers.25.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 0 + }, + { + "name": "model.layers.25.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 6291456 + }, + { + "name": "model.layers.26.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 7077888 + }, + { + "name": "model.layers.26.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 7084032 + }, + { + "name": "model.layers.26.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 11802624 + }, + { + "name": "model.layers.26.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 21239808 + }, + { + "name": "model.layers.26.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 21245952 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 23605248 + }, + { + "name": "model.layers.26.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 29896704 + }, + { + "name": "model.layers.27.input_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 30683136 + } + ], + "md5sum": "556ecbf627fe71befbd3cbc4fca9c223" + }, + { + "dataPath": "params_shard_99.bin", + "format": "raw-shard", + "nbytes": 75497472, + "records": [ + { + "name": "model.layers.27.mlp.gate_up_proj.q_weight", + "shape": [ + 384, + 49152 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 75497472, + "byteOffset": 0 + } + ], + "md5sum": "f12eb6c39de5424dd4adedd9657e5d5e" + }, + { + "dataPath": "params_shard_100.bin", + "format": "raw-shard", + "nbytes": 33036288, + "records": [ + { + "name": "model.layers.27.mlp.down_proj.q_scale", + "shape": [ + 768, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 4718592, + "byteOffset": 0 + }, + { + "name": "model.layers.27.mlp.gate_up_proj.q_scale", + "shape": [ + 96, + 49152 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 9437184, + "byteOffset": 4718592 + }, + { + "name": "model.layers.27.post_attention_layernorm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 14155776 + }, + { + "name": "model.layers.27.self_attn.qkv_proj.q_weight", + "shape": [ + 384, + 12288 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 18874368, + "byteOffset": 14161920 + } + ], + "md5sum": "a69aa623d966a661e7e77a24d84d954b" + }, + { + "dataPath": "params_shard_101.bin", + "format": "raw-shard", + "nbytes": 9443328, + "records": [ + { + "name": "model.layers.27.self_attn.qkv_proj.q_scale", + "shape": [ + 96, + 12288 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 2359296, + "byteOffset": 0 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_weight", + "shape": [ + 512, + 3072 + ], + "dtype": "uint32", + "format": "f32-to-bf16", + "nbytes": 6291456, + "byteOffset": 2359296 + }, + { + "name": "model.layers.27.self_attn.o_proj.q_scale", + "shape": [ + 128, + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 786432, + "byteOffset": 8650752 + }, + { + "name": "model.norm.weight", + "shape": [ + 3072 + ], + "dtype": "float16", + "format": "f32-to-bf16", + "nbytes": 6144, + "byteOffset": 9437184 + } + ], + "md5sum": "7cbf7bd122b10968c893e9eb861abf72" + } + ] +} \ No newline at end of file