{ "metadata": { "ParamSize": 518, "ParamBytes": 1563781120.0, "BitsPerParam": 4.506794107285474 }, "records": [ { "dataPath": "params_shard_0.bin", "format": "raw-shard", "nbytes": 64552960, "records": [ { "name": "gpt_neox.embed_in.q_weight", "shape": [ 50432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 64552960, "byteOffset": 0 } ], "md5sum": "a1e5a97b805954a1bf21fa5894c1d0f3" }, { "dataPath": "params_shard_1.bin", "format": "raw-shard", "nbytes": 22855680, "records": [ { "name": "gpt_neox.embed_in.q_scale", "shape": [ 50432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8069120, "byteOffset": 0 }, { "name": "gpt_neox.layers.0.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 8069120 }, { "name": "gpt_neox.layers.0.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 8074240 }, { "name": "gpt_neox.layers.0.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 8079360 }, { "name": "gpt_neox.layers.0.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 8084480 }, { "name": "gpt_neox.layers.0.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 8089600 }, { "name": "gpt_neox.layers.0.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 17920000 }, { "name": "gpt_neox.layers.0.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 19148800 }, { "name": "gpt_neox.layers.0.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 19164160 }, { "name": "gpt_neox.layers.0.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 22440960 }, { "name": "gpt_neox.layers.0.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 22850560 } ], "md5sum": "397aecdda553b0fd5e8e21c046ffc28f" }, { "dataPath": "params_shard_2.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.0.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.0.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.0.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.0.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.0.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.0.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.1.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.1.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.1.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.1.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "29c763ddaf1851418476e78874d42734" }, { "dataPath": "params_shard_3.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.1.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.1.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.1.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.1.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.1.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.1.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.1.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.1.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.1.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "de34e9206ec7c2851be3392728fd5b98" }, { "dataPath": "params_shard_4.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.1.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.1.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.1.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.2.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.2.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.2.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.2.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.2.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.2.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.2.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.2.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.2.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.2.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "45e45fda99df6d431b13c38b987bf0a7" }, { "dataPath": "params_shard_5.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.2.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.2.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.2.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.2.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.2.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.2.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.3.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.3.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.3.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.3.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "208b1b14adadd35dbaa2a6952ee9962b" }, { "dataPath": "params_shard_6.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.3.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.3.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.3.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.3.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.3.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.3.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.3.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.3.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.3.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "97f99579c8083568f3274f0bfc245b68" }, { "dataPath": "params_shard_7.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.3.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.3.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.3.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.4.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.4.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.4.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.4.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.4.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.4.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.4.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.4.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.4.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.4.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "a7087a9561e7ea1496559a0580146d40" }, { "dataPath": "params_shard_8.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.4.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.4.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.4.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.4.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.4.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.4.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.5.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.5.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.5.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.5.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "f5e671e03a30ec3ca1b258f19ccf3f22" }, { "dataPath": "params_shard_9.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.5.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.5.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.5.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.5.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.5.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.5.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.5.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.5.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.5.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "6baee38d3fcb208bce788c8732dce16b" }, { "dataPath": "params_shard_10.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.5.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.5.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.5.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.6.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.6.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.6.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.6.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.6.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.6.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.6.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.6.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.6.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.6.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "a5e6a99184e7036911ab4f85bee80fe9" }, { "dataPath": "params_shard_11.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.6.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.6.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.6.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.6.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.6.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.6.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.7.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.7.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.7.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.7.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "539e0740360a120b3a73fa93e6949056" }, { "dataPath": "params_shard_12.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.7.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.7.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.7.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.7.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.7.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.7.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.7.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.7.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.7.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "520be8c4dca8b5375e049b3ad60d5468" }, { "dataPath": "params_shard_13.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.7.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.7.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.7.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.8.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.8.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.8.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.8.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.8.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.8.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.8.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.8.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.8.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.8.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "65d17d53dd8b7e11d919f4aad3f5b379" }, { "dataPath": "params_shard_14.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.8.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.8.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.8.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.8.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.8.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.8.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.9.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.9.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.9.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.9.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "a406cf5d26110368448dbdc3f9b96fd6" }, { "dataPath": "params_shard_15.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.9.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.9.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.9.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.9.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.9.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.9.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.9.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.9.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.9.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "62d3a3610e747e2ddec5e17d3abec711" }, { "dataPath": "params_shard_16.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.9.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.9.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.9.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.10.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.10.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.10.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.10.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.10.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.10.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.10.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.10.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.10.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.10.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "c49a2b684f7501511759c1b4e9b2662c" }, { "dataPath": "params_shard_17.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.10.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.10.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.10.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.10.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.10.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.10.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.11.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.11.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.11.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.11.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "a843865483d64fb455d9d0eb0fd85dc4" }, { "dataPath": "params_shard_18.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.11.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.11.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.11.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.11.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.11.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.11.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.11.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.11.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.11.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "fff386fb7649569f93c73fbd5c269ae7" }, { "dataPath": "params_shard_19.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.11.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.11.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.11.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.12.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.12.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.12.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.12.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.12.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.12.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.12.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.12.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.12.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.12.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "da4b6c3cfcc0cfeb74f01eeef780f5d0" }, { "dataPath": "params_shard_20.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.12.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.12.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.12.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.12.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.12.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.12.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.13.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.13.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.13.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.13.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "2ca94feeb55d05f40787d3b88e1e1bbd" }, { "dataPath": "params_shard_21.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.13.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.13.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.13.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.13.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.13.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.13.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.13.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.13.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.13.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "504c0a2332fb0f2752bfd1adfbb9d9b9" }, { "dataPath": "params_shard_22.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.13.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.13.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.13.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.14.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.14.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.14.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.14.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.14.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.14.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.14.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.14.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.14.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.14.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "2040d26d9d291c43fb3a7519d44fce28" }, { "dataPath": "params_shard_23.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.14.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.14.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.14.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.14.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.14.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.14.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.15.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.15.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.15.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.15.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "80b9073461cce961eb4c12d08193d5a7" }, { "dataPath": "params_shard_24.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.15.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.15.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.15.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.15.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.15.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.15.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.15.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.15.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.15.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "bbb68d2fe44eb5499f1e69c1ca949071" }, { "dataPath": "params_shard_25.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.15.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.15.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.15.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.16.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.16.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.16.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.16.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.16.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.16.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.16.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.16.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.16.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.16.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "1c957a8c6fcb6c186426a320bfe1a5f3" }, { "dataPath": "params_shard_26.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.16.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.16.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.16.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.16.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.16.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.16.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.17.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.17.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.17.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.17.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "b943f3be1ffd53a55f3c20f930bff863" }, { "dataPath": "params_shard_27.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.17.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.17.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.17.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.17.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.17.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.17.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.17.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.17.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.17.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "5ed9117031958a9274c7af99e869713a" }, { "dataPath": "params_shard_28.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.17.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.17.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.17.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.18.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.18.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.18.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.18.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.18.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.18.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.18.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.18.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.18.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.18.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "64d4c61488bdb328641d24dcd1d67fff" }, { "dataPath": "params_shard_29.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.18.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.18.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.18.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.18.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.18.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.18.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.19.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.19.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.19.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.19.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "b76073d35df66ef0c4f895a289e26dd6" }, { "dataPath": "params_shard_30.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.19.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.19.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.19.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.19.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.19.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.19.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.19.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.19.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.19.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "7fb88761476fdd62a2375d2140371473" }, { "dataPath": "params_shard_31.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.19.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.19.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.19.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.20.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.20.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.20.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.20.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.20.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.20.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.20.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.20.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.20.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.20.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "82d6389ed115b5c9e15aa0c855c89069" }, { "dataPath": "params_shard_32.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.20.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.20.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.20.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.20.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.20.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.20.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.21.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.21.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.21.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.21.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "ec87278abcbd19366700d852b0e4b57d" }, { "dataPath": "params_shard_33.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.21.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.21.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.21.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.21.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.21.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.21.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.21.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.21.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.21.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "077c44888910807f9686298807c46b06" }, { "dataPath": "params_shard_34.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.21.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.21.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.21.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.22.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.22.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.22.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.22.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.22.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.22.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.22.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.22.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.22.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.22.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "e0a5640327127d6c08ed8956bf87e1b1" }, { "dataPath": "params_shard_35.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.22.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.22.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.22.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.22.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.22.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.22.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.23.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.23.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.23.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.23.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "a66f59f242c08c63b46ac9f416351b91" }, { "dataPath": "params_shard_36.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.23.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.23.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.23.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.23.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.23.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.23.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.23.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.23.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.23.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "fdb91ab9a99d0b59f0d34abe57bc666e" }, { "dataPath": "params_shard_37.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.23.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.23.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.23.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.24.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.24.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.24.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.24.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.24.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.24.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.24.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.24.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.24.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.24.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "fab8bb3909bb27a991c4d3734305a2ba" }, { "dataPath": "params_shard_38.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.24.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.24.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.24.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.24.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.24.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.24.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.25.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.25.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.25.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.25.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "bcac0181365628d0cd4f547ec2507ea2" }, { "dataPath": "params_shard_39.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.25.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.25.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.25.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.25.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.25.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.25.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.25.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.25.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.25.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "4b27532d0f518caa0942c91d815ac5ff" }, { "dataPath": "params_shard_40.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.25.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.25.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.25.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.26.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.26.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.26.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.26.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.26.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.26.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.26.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.26.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.26.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.26.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "cde4d8c7e23fa419cd5d84ff2e6d6b1e" }, { "dataPath": "params_shard_41.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.26.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.26.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.26.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.26.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.26.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.26.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.27.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.27.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.27.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.27.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "8af216195a504fb290d6113c4bd84648" }, { "dataPath": "params_shard_42.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.27.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.27.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.27.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.27.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.27.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.27.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.27.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.27.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.27.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "9a6b53d582c45e58b5fac9c7eee43640" }, { "dataPath": "params_shard_43.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.27.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.27.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.27.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.28.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.28.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.28.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.28.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.28.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.28.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.28.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.28.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.28.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.28.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "3c64e8b7a03aa6216df97e2f241e42b3" }, { "dataPath": "params_shard_44.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.28.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.28.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.28.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.28.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.28.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.28.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.29.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.29.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.29.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.29.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "d1089ee5048be76261846a03fe2cfab6" }, { "dataPath": "params_shard_45.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.29.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.29.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.29.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.29.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.29.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.29.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.29.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.29.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.29.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "bd62fa056b9572490c59dd176ad96b09" }, { "dataPath": "params_shard_46.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.29.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.29.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.29.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.30.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.layers.30.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "gpt_neox.layers.30.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.30.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.30.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 14771200 }, { "name": "gpt_neox.layers.30.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 24601600 }, { "name": "gpt_neox.layers.30.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 25830400 }, { "name": "gpt_neox.layers.30.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 25845760 }, { "name": "gpt_neox.layers.30.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 29122560 }, { "name": "gpt_neox.layers.30.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "4f30fd59aa23cdac1cc16774557a4b08" }, { "dataPath": "params_shard_47.bin", "format": "raw-shard", "nbytes": 29537280, "records": [ { "name": "gpt_neox.layers.30.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.30.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.30.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 14745600 }, { "name": "gpt_neox.layers.30.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.30.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.30.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29511680 }, { "name": "gpt_neox.layers.31.input_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29516800 }, { "name": "gpt_neox.layers.31.input_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29521920 }, { "name": "gpt_neox.layers.31.post_attention_layernorm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29527040 }, { "name": "gpt_neox.layers.31.post_attention_layernorm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 29532160 } ], "md5sum": "b0778e66db9e05b929741416e66a0a3c" }, { "dataPath": "params_shard_48.bin", "format": "raw-shard", "nbytes": 29532160, "records": [ { "name": "gpt_neox.layers.31.attention.query_key_value.q_weight", "shape": [ 7680, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 9830400, "byteOffset": 0 }, { "name": "gpt_neox.layers.31.attention.query_key_value.q_scale", "shape": [ 7680, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1228800, "byteOffset": 9830400 }, { "name": "gpt_neox.layers.31.attention.query_key_value.bias", "shape": [ 7680 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 15360, "byteOffset": 11059200 }, { "name": "gpt_neox.layers.31.attention.dense.q_weight", "shape": [ 2560, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 3276800, "byteOffset": 11074560 }, { "name": "gpt_neox.layers.31.attention.dense.q_scale", "shape": [ 2560, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 409600, "byteOffset": 14351360 }, { "name": "gpt_neox.layers.31.attention.dense.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14760960 }, { "name": "gpt_neox.layers.31.mlp.dense_h_to_4h.q_weight", "shape": [ 10240, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 14766080 }, { "name": "gpt_neox.layers.31.mlp.dense_h_to_4h.q_scale", "shape": [ 10240, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 27873280 }, { "name": "gpt_neox.layers.31.mlp.dense_h_to_4h.bias", "shape": [ 10240 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 20480, "byteOffset": 29511680 } ], "md5sum": "cc08bfb065efe7f85a7e23d7f5457cec" }, { "dataPath": "params_shard_49.bin", "format": "raw-shard", "nbytes": 64552960, "records": [ { "name": "embed_out.q_weight", "shape": [ 50432, 320 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 64552960, "byteOffset": 0 } ], "md5sum": "3c52aa30ed63b1a12ad555cd4160c06b" }, { "dataPath": "params_shard_50.bin", "format": "raw-shard", "nbytes": 22830080, "records": [ { "name": "gpt_neox.layers.31.mlp.dense_4h_to_h.q_weight", "shape": [ 2560, 1280 ], "dtype": "uint32", "format": "f32-to-bf16", "nbytes": 13107200, "byteOffset": 0 }, { "name": "gpt_neox.layers.31.mlp.dense_4h_to_h.q_scale", "shape": [ 2560, 320 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 1638400, "byteOffset": 13107200 }, { "name": "gpt_neox.layers.31.mlp.dense_4h_to_h.bias", "shape": [ 2560 ], "dtype": "float32", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14745600 }, { "name": "gpt_neox.final_layer_norm.weight", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14750720 }, { "name": "gpt_neox.final_layer_norm.bias", "shape": [ 2560 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 5120, "byteOffset": 14755840 }, { "name": "embed_out.q_scale", "shape": [ 50432, 80 ], "dtype": "float16", "format": "f32-to-bf16", "nbytes": 8069120, "byteOffset": 14760960 } ], "md5sum": "c0816787ad914e6010834840239a0960" } ] }