{ "schemaVersion": 2, "format": "revis-xlm-roberta-e5-mlx-native-q8", "source": "intfloat/multilingual-e5-large", "baseWeights": "weights.00.safetensors", "quantizedWeights": "weights.00.safetensors", "quantization": { "type": "mlx-native-affine", "bits": 8, "group_size": 64, "mode": "affine", "quantizedTensorSuffix": ".qweight", "scaleTensorSuffix": ".scales", "biasTensorSuffix": ".biases", "linear": "mx.quantized_matmul(x, qweight, scales=scales, biases=biases, transpose=True, group_size=group_size, bits=bits, mode=mode)", "embedding": "mx.dequantize(qweight[ids], scales=scales[ids], biases=biases[ids], group_size=group_size, bits=bits, mode=mode)" }, "quantized": [ { "name": "embeddings.position_embeddings.weight", "shape": [ 514, 1024 ], "dtype": "float16", "qweight": "embeddings.position_embeddings.weight.qweight", "scales": "embeddings.position_embeddings.weight.scales", "biases": "embeddings.position_embeddings.weight.biases" }, { "name": "embeddings.token_type_embeddings.weight", "shape": [ 1, 1024 ], "dtype": "float16", "qweight": "embeddings.token_type_embeddings.weight.qweight", "scales": "embeddings.token_type_embeddings.weight.scales", "biases": "embeddings.token_type_embeddings.weight.biases" }, { "name": "embeddings.word_embeddings.weight", "shape": [ 250002, 1024 ], "dtype": "float16", "qweight": "embeddings.word_embeddings.weight.qweight", "scales": "embeddings.word_embeddings.weight.scales", "biases": "embeddings.word_embeddings.weight.biases" }, { "name": "encoder.layer.0.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.0.attention.output.dense.weight.qweight", "scales": "encoder.layer.0.attention.output.dense.weight.scales", "biases": "encoder.layer.0.attention.output.dense.weight.biases" }, { "name": "encoder.layer.0.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.0.attention.self.key.weight.qweight", "scales": "encoder.layer.0.attention.self.key.weight.scales", "biases": "encoder.layer.0.attention.self.key.weight.biases" }, { "name": "encoder.layer.0.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.0.attention.self.query.weight.qweight", "scales": "encoder.layer.0.attention.self.query.weight.scales", "biases": "encoder.layer.0.attention.self.query.weight.biases" }, { "name": "encoder.layer.0.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.0.attention.self.value.weight.qweight", "scales": "encoder.layer.0.attention.self.value.weight.scales", "biases": "encoder.layer.0.attention.self.value.weight.biases" }, { "name": "encoder.layer.0.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.0.intermediate.dense.weight.qweight", "scales": "encoder.layer.0.intermediate.dense.weight.scales", "biases": "encoder.layer.0.intermediate.dense.weight.biases" }, { "name": "encoder.layer.0.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.0.output.dense.weight.qweight", "scales": "encoder.layer.0.output.dense.weight.scales", "biases": "encoder.layer.0.output.dense.weight.biases" }, { "name": "encoder.layer.1.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.1.attention.output.dense.weight.qweight", "scales": "encoder.layer.1.attention.output.dense.weight.scales", "biases": "encoder.layer.1.attention.output.dense.weight.biases" }, { "name": "encoder.layer.1.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.1.attention.self.key.weight.qweight", "scales": "encoder.layer.1.attention.self.key.weight.scales", "biases": "encoder.layer.1.attention.self.key.weight.biases" }, { "name": "encoder.layer.1.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.1.attention.self.query.weight.qweight", "scales": "encoder.layer.1.attention.self.query.weight.scales", "biases": "encoder.layer.1.attention.self.query.weight.biases" }, { "name": "encoder.layer.1.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.1.attention.self.value.weight.qweight", "scales": "encoder.layer.1.attention.self.value.weight.scales", "biases": "encoder.layer.1.attention.self.value.weight.biases" }, { "name": "encoder.layer.1.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.1.intermediate.dense.weight.qweight", "scales": "encoder.layer.1.intermediate.dense.weight.scales", "biases": "encoder.layer.1.intermediate.dense.weight.biases" }, { "name": "encoder.layer.1.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.1.output.dense.weight.qweight", "scales": "encoder.layer.1.output.dense.weight.scales", "biases": "encoder.layer.1.output.dense.weight.biases" }, { "name": "encoder.layer.10.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.10.attention.output.dense.weight.qweight", "scales": "encoder.layer.10.attention.output.dense.weight.scales", "biases": "encoder.layer.10.attention.output.dense.weight.biases" }, { "name": "encoder.layer.10.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.10.attention.self.key.weight.qweight", "scales": "encoder.layer.10.attention.self.key.weight.scales", "biases": "encoder.layer.10.attention.self.key.weight.biases" }, { "name": "encoder.layer.10.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.10.attention.self.query.weight.qweight", "scales": "encoder.layer.10.attention.self.query.weight.scales", "biases": "encoder.layer.10.attention.self.query.weight.biases" }, { "name": "encoder.layer.10.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.10.attention.self.value.weight.qweight", "scales": "encoder.layer.10.attention.self.value.weight.scales", "biases": "encoder.layer.10.attention.self.value.weight.biases" }, { "name": "encoder.layer.10.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.10.intermediate.dense.weight.qweight", "scales": "encoder.layer.10.intermediate.dense.weight.scales", "biases": "encoder.layer.10.intermediate.dense.weight.biases" }, { "name": "encoder.layer.10.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.10.output.dense.weight.qweight", "scales": "encoder.layer.10.output.dense.weight.scales", "biases": "encoder.layer.10.output.dense.weight.biases" }, { "name": "encoder.layer.11.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.11.attention.output.dense.weight.qweight", "scales": "encoder.layer.11.attention.output.dense.weight.scales", "biases": "encoder.layer.11.attention.output.dense.weight.biases" }, { "name": "encoder.layer.11.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.11.attention.self.key.weight.qweight", "scales": "encoder.layer.11.attention.self.key.weight.scales", "biases": "encoder.layer.11.attention.self.key.weight.biases" }, { "name": "encoder.layer.11.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.11.attention.self.query.weight.qweight", "scales": "encoder.layer.11.attention.self.query.weight.scales", "biases": "encoder.layer.11.attention.self.query.weight.biases" }, { "name": "encoder.layer.11.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.11.attention.self.value.weight.qweight", "scales": "encoder.layer.11.attention.self.value.weight.scales", "biases": "encoder.layer.11.attention.self.value.weight.biases" }, { "name": "encoder.layer.11.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.11.intermediate.dense.weight.qweight", "scales": "encoder.layer.11.intermediate.dense.weight.scales", "biases": "encoder.layer.11.intermediate.dense.weight.biases" }, { "name": "encoder.layer.11.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.11.output.dense.weight.qweight", "scales": "encoder.layer.11.output.dense.weight.scales", "biases": "encoder.layer.11.output.dense.weight.biases" }, { "name": "encoder.layer.12.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.12.attention.output.dense.weight.qweight", "scales": "encoder.layer.12.attention.output.dense.weight.scales", "biases": "encoder.layer.12.attention.output.dense.weight.biases" }, { "name": "encoder.layer.12.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.12.attention.self.key.weight.qweight", "scales": "encoder.layer.12.attention.self.key.weight.scales", "biases": "encoder.layer.12.attention.self.key.weight.biases" }, { "name": "encoder.layer.12.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.12.attention.self.query.weight.qweight", "scales": "encoder.layer.12.attention.self.query.weight.scales", "biases": "encoder.layer.12.attention.self.query.weight.biases" }, { "name": "encoder.layer.12.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.12.attention.self.value.weight.qweight", "scales": "encoder.layer.12.attention.self.value.weight.scales", "biases": "encoder.layer.12.attention.self.value.weight.biases" }, { "name": "encoder.layer.12.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.12.intermediate.dense.weight.qweight", "scales": "encoder.layer.12.intermediate.dense.weight.scales", "biases": "encoder.layer.12.intermediate.dense.weight.biases" }, { "name": "encoder.layer.12.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.12.output.dense.weight.qweight", "scales": "encoder.layer.12.output.dense.weight.scales", "biases": "encoder.layer.12.output.dense.weight.biases" }, { "name": "encoder.layer.13.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.13.attention.output.dense.weight.qweight", "scales": "encoder.layer.13.attention.output.dense.weight.scales", "biases": "encoder.layer.13.attention.output.dense.weight.biases" }, { "name": "encoder.layer.13.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.13.attention.self.key.weight.qweight", "scales": "encoder.layer.13.attention.self.key.weight.scales", "biases": "encoder.layer.13.attention.self.key.weight.biases" }, { "name": "encoder.layer.13.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.13.attention.self.query.weight.qweight", "scales": "encoder.layer.13.attention.self.query.weight.scales", "biases": "encoder.layer.13.attention.self.query.weight.biases" }, { "name": "encoder.layer.13.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.13.attention.self.value.weight.qweight", "scales": "encoder.layer.13.attention.self.value.weight.scales", "biases": "encoder.layer.13.attention.self.value.weight.biases" }, { "name": "encoder.layer.13.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.13.intermediate.dense.weight.qweight", "scales": "encoder.layer.13.intermediate.dense.weight.scales", "biases": "encoder.layer.13.intermediate.dense.weight.biases" }, { "name": "encoder.layer.13.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.13.output.dense.weight.qweight", "scales": "encoder.layer.13.output.dense.weight.scales", "biases": "encoder.layer.13.output.dense.weight.biases" }, { "name": "encoder.layer.14.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.14.attention.output.dense.weight.qweight", "scales": "encoder.layer.14.attention.output.dense.weight.scales", "biases": "encoder.layer.14.attention.output.dense.weight.biases" }, { "name": "encoder.layer.14.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.14.attention.self.key.weight.qweight", "scales": "encoder.layer.14.attention.self.key.weight.scales", "biases": "encoder.layer.14.attention.self.key.weight.biases" }, { "name": "encoder.layer.14.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.14.attention.self.query.weight.qweight", "scales": "encoder.layer.14.attention.self.query.weight.scales", "biases": "encoder.layer.14.attention.self.query.weight.biases" }, { "name": "encoder.layer.14.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.14.attention.self.value.weight.qweight", "scales": "encoder.layer.14.attention.self.value.weight.scales", "biases": "encoder.layer.14.attention.self.value.weight.biases" }, { "name": "encoder.layer.14.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.14.intermediate.dense.weight.qweight", "scales": "encoder.layer.14.intermediate.dense.weight.scales", "biases": "encoder.layer.14.intermediate.dense.weight.biases" }, { "name": "encoder.layer.14.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.14.output.dense.weight.qweight", "scales": "encoder.layer.14.output.dense.weight.scales", "biases": "encoder.layer.14.output.dense.weight.biases" }, { "name": "encoder.layer.15.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.15.attention.output.dense.weight.qweight", "scales": "encoder.layer.15.attention.output.dense.weight.scales", "biases": "encoder.layer.15.attention.output.dense.weight.biases" }, { "name": "encoder.layer.15.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.15.attention.self.key.weight.qweight", "scales": "encoder.layer.15.attention.self.key.weight.scales", "biases": "encoder.layer.15.attention.self.key.weight.biases" }, { "name": "encoder.layer.15.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.15.attention.self.query.weight.qweight", "scales": "encoder.layer.15.attention.self.query.weight.scales", "biases": "encoder.layer.15.attention.self.query.weight.biases" }, { "name": "encoder.layer.15.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.15.attention.self.value.weight.qweight", "scales": "encoder.layer.15.attention.self.value.weight.scales", "biases": "encoder.layer.15.attention.self.value.weight.biases" }, { "name": "encoder.layer.15.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.15.intermediate.dense.weight.qweight", "scales": "encoder.layer.15.intermediate.dense.weight.scales", "biases": "encoder.layer.15.intermediate.dense.weight.biases" }, { "name": "encoder.layer.15.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.15.output.dense.weight.qweight", "scales": "encoder.layer.15.output.dense.weight.scales", "biases": "encoder.layer.15.output.dense.weight.biases" }, { "name": "encoder.layer.16.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.16.attention.output.dense.weight.qweight", "scales": "encoder.layer.16.attention.output.dense.weight.scales", "biases": "encoder.layer.16.attention.output.dense.weight.biases" }, { "name": "encoder.layer.16.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.16.attention.self.key.weight.qweight", "scales": "encoder.layer.16.attention.self.key.weight.scales", "biases": "encoder.layer.16.attention.self.key.weight.biases" }, { "name": "encoder.layer.16.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.16.attention.self.query.weight.qweight", "scales": "encoder.layer.16.attention.self.query.weight.scales", "biases": "encoder.layer.16.attention.self.query.weight.biases" }, { "name": "encoder.layer.16.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.16.attention.self.value.weight.qweight", "scales": "encoder.layer.16.attention.self.value.weight.scales", "biases": "encoder.layer.16.attention.self.value.weight.biases" }, { "name": "encoder.layer.16.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.16.intermediate.dense.weight.qweight", "scales": "encoder.layer.16.intermediate.dense.weight.scales", "biases": "encoder.layer.16.intermediate.dense.weight.biases" }, { "name": "encoder.layer.16.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.16.output.dense.weight.qweight", "scales": "encoder.layer.16.output.dense.weight.scales", "biases": "encoder.layer.16.output.dense.weight.biases" }, { "name": "encoder.layer.17.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.17.attention.output.dense.weight.qweight", "scales": "encoder.layer.17.attention.output.dense.weight.scales", "biases": "encoder.layer.17.attention.output.dense.weight.biases" }, { "name": "encoder.layer.17.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.17.attention.self.key.weight.qweight", "scales": "encoder.layer.17.attention.self.key.weight.scales", "biases": "encoder.layer.17.attention.self.key.weight.biases" }, { "name": "encoder.layer.17.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.17.attention.self.query.weight.qweight", "scales": "encoder.layer.17.attention.self.query.weight.scales", "biases": "encoder.layer.17.attention.self.query.weight.biases" }, { "name": "encoder.layer.17.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.17.attention.self.value.weight.qweight", "scales": "encoder.layer.17.attention.self.value.weight.scales", "biases": "encoder.layer.17.attention.self.value.weight.biases" }, { "name": "encoder.layer.17.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.17.intermediate.dense.weight.qweight", "scales": "encoder.layer.17.intermediate.dense.weight.scales", "biases": "encoder.layer.17.intermediate.dense.weight.biases" }, { "name": "encoder.layer.17.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.17.output.dense.weight.qweight", "scales": "encoder.layer.17.output.dense.weight.scales", "biases": "encoder.layer.17.output.dense.weight.biases" }, { "name": "encoder.layer.18.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.18.attention.output.dense.weight.qweight", "scales": "encoder.layer.18.attention.output.dense.weight.scales", "biases": "encoder.layer.18.attention.output.dense.weight.biases" }, { "name": "encoder.layer.18.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.18.attention.self.key.weight.qweight", "scales": "encoder.layer.18.attention.self.key.weight.scales", "biases": "encoder.layer.18.attention.self.key.weight.biases" }, { "name": "encoder.layer.18.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.18.attention.self.query.weight.qweight", "scales": "encoder.layer.18.attention.self.query.weight.scales", "biases": "encoder.layer.18.attention.self.query.weight.biases" }, { "name": "encoder.layer.18.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.18.attention.self.value.weight.qweight", "scales": "encoder.layer.18.attention.self.value.weight.scales", "biases": "encoder.layer.18.attention.self.value.weight.biases" }, { "name": "encoder.layer.18.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.18.intermediate.dense.weight.qweight", "scales": "encoder.layer.18.intermediate.dense.weight.scales", "biases": "encoder.layer.18.intermediate.dense.weight.biases" }, { "name": "encoder.layer.18.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.18.output.dense.weight.qweight", "scales": "encoder.layer.18.output.dense.weight.scales", "biases": "encoder.layer.18.output.dense.weight.biases" }, { "name": "encoder.layer.19.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.19.attention.output.dense.weight.qweight", "scales": "encoder.layer.19.attention.output.dense.weight.scales", "biases": "encoder.layer.19.attention.output.dense.weight.biases" }, { "name": "encoder.layer.19.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.19.attention.self.key.weight.qweight", "scales": "encoder.layer.19.attention.self.key.weight.scales", "biases": "encoder.layer.19.attention.self.key.weight.biases" }, { "name": "encoder.layer.19.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.19.attention.self.query.weight.qweight", "scales": "encoder.layer.19.attention.self.query.weight.scales", "biases": "encoder.layer.19.attention.self.query.weight.biases" }, { "name": "encoder.layer.19.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.19.attention.self.value.weight.qweight", "scales": "encoder.layer.19.attention.self.value.weight.scales", "biases": "encoder.layer.19.attention.self.value.weight.biases" }, { "name": "encoder.layer.19.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.19.intermediate.dense.weight.qweight", "scales": "encoder.layer.19.intermediate.dense.weight.scales", "biases": "encoder.layer.19.intermediate.dense.weight.biases" }, { "name": "encoder.layer.19.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.19.output.dense.weight.qweight", "scales": "encoder.layer.19.output.dense.weight.scales", "biases": "encoder.layer.19.output.dense.weight.biases" }, { "name": "encoder.layer.2.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.2.attention.output.dense.weight.qweight", "scales": "encoder.layer.2.attention.output.dense.weight.scales", "biases": "encoder.layer.2.attention.output.dense.weight.biases" }, { "name": "encoder.layer.2.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.2.attention.self.key.weight.qweight", "scales": "encoder.layer.2.attention.self.key.weight.scales", "biases": "encoder.layer.2.attention.self.key.weight.biases" }, { "name": "encoder.layer.2.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.2.attention.self.query.weight.qweight", "scales": "encoder.layer.2.attention.self.query.weight.scales", "biases": "encoder.layer.2.attention.self.query.weight.biases" }, { "name": "encoder.layer.2.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.2.attention.self.value.weight.qweight", "scales": "encoder.layer.2.attention.self.value.weight.scales", "biases": "encoder.layer.2.attention.self.value.weight.biases" }, { "name": "encoder.layer.2.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.2.intermediate.dense.weight.qweight", "scales": "encoder.layer.2.intermediate.dense.weight.scales", "biases": "encoder.layer.2.intermediate.dense.weight.biases" }, { "name": "encoder.layer.2.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.2.output.dense.weight.qweight", "scales": "encoder.layer.2.output.dense.weight.scales", "biases": "encoder.layer.2.output.dense.weight.biases" }, { "name": "encoder.layer.20.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.20.attention.output.dense.weight.qweight", "scales": "encoder.layer.20.attention.output.dense.weight.scales", "biases": "encoder.layer.20.attention.output.dense.weight.biases" }, { "name": "encoder.layer.20.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.20.attention.self.key.weight.qweight", "scales": "encoder.layer.20.attention.self.key.weight.scales", "biases": "encoder.layer.20.attention.self.key.weight.biases" }, { "name": "encoder.layer.20.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.20.attention.self.query.weight.qweight", "scales": "encoder.layer.20.attention.self.query.weight.scales", "biases": "encoder.layer.20.attention.self.query.weight.biases" }, { "name": "encoder.layer.20.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.20.attention.self.value.weight.qweight", "scales": "encoder.layer.20.attention.self.value.weight.scales", "biases": "encoder.layer.20.attention.self.value.weight.biases" }, { "name": "encoder.layer.20.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.20.intermediate.dense.weight.qweight", "scales": "encoder.layer.20.intermediate.dense.weight.scales", "biases": "encoder.layer.20.intermediate.dense.weight.biases" }, { "name": "encoder.layer.20.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.20.output.dense.weight.qweight", "scales": "encoder.layer.20.output.dense.weight.scales", "biases": "encoder.layer.20.output.dense.weight.biases" }, { "name": "encoder.layer.21.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.21.attention.output.dense.weight.qweight", "scales": "encoder.layer.21.attention.output.dense.weight.scales", "biases": "encoder.layer.21.attention.output.dense.weight.biases" }, { "name": "encoder.layer.21.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.21.attention.self.key.weight.qweight", "scales": "encoder.layer.21.attention.self.key.weight.scales", "biases": "encoder.layer.21.attention.self.key.weight.biases" }, { "name": "encoder.layer.21.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.21.attention.self.query.weight.qweight", "scales": "encoder.layer.21.attention.self.query.weight.scales", "biases": "encoder.layer.21.attention.self.query.weight.biases" }, { "name": "encoder.layer.21.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.21.attention.self.value.weight.qweight", "scales": "encoder.layer.21.attention.self.value.weight.scales", "biases": "encoder.layer.21.attention.self.value.weight.biases" }, { "name": "encoder.layer.21.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.21.intermediate.dense.weight.qweight", "scales": "encoder.layer.21.intermediate.dense.weight.scales", "biases": "encoder.layer.21.intermediate.dense.weight.biases" }, { "name": "encoder.layer.21.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.21.output.dense.weight.qweight", "scales": "encoder.layer.21.output.dense.weight.scales", "biases": "encoder.layer.21.output.dense.weight.biases" }, { "name": "encoder.layer.22.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.22.attention.output.dense.weight.qweight", "scales": "encoder.layer.22.attention.output.dense.weight.scales", "biases": "encoder.layer.22.attention.output.dense.weight.biases" }, { "name": "encoder.layer.22.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.22.attention.self.key.weight.qweight", "scales": "encoder.layer.22.attention.self.key.weight.scales", "biases": "encoder.layer.22.attention.self.key.weight.biases" }, { "name": "encoder.layer.22.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.22.attention.self.query.weight.qweight", "scales": "encoder.layer.22.attention.self.query.weight.scales", "biases": "encoder.layer.22.attention.self.query.weight.biases" }, { "name": "encoder.layer.22.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.22.attention.self.value.weight.qweight", "scales": "encoder.layer.22.attention.self.value.weight.scales", "biases": "encoder.layer.22.attention.self.value.weight.biases" }, { "name": "encoder.layer.22.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.22.intermediate.dense.weight.qweight", "scales": "encoder.layer.22.intermediate.dense.weight.scales", "biases": "encoder.layer.22.intermediate.dense.weight.biases" }, { "name": "encoder.layer.22.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.22.output.dense.weight.qweight", "scales": "encoder.layer.22.output.dense.weight.scales", "biases": "encoder.layer.22.output.dense.weight.biases" }, { "name": "encoder.layer.23.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.23.attention.output.dense.weight.qweight", "scales": "encoder.layer.23.attention.output.dense.weight.scales", "biases": "encoder.layer.23.attention.output.dense.weight.biases" }, { "name": "encoder.layer.23.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.23.attention.self.key.weight.qweight", "scales": "encoder.layer.23.attention.self.key.weight.scales", "biases": "encoder.layer.23.attention.self.key.weight.biases" }, { "name": "encoder.layer.23.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.23.attention.self.query.weight.qweight", "scales": "encoder.layer.23.attention.self.query.weight.scales", "biases": "encoder.layer.23.attention.self.query.weight.biases" }, { "name": "encoder.layer.23.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.23.attention.self.value.weight.qweight", "scales": "encoder.layer.23.attention.self.value.weight.scales", "biases": "encoder.layer.23.attention.self.value.weight.biases" }, { "name": "encoder.layer.23.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.23.intermediate.dense.weight.qweight", "scales": "encoder.layer.23.intermediate.dense.weight.scales", "biases": "encoder.layer.23.intermediate.dense.weight.biases" }, { "name": "encoder.layer.23.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.23.output.dense.weight.qweight", "scales": "encoder.layer.23.output.dense.weight.scales", "biases": "encoder.layer.23.output.dense.weight.biases" }, { "name": "encoder.layer.3.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.3.attention.output.dense.weight.qweight", "scales": "encoder.layer.3.attention.output.dense.weight.scales", "biases": "encoder.layer.3.attention.output.dense.weight.biases" }, { "name": "encoder.layer.3.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.3.attention.self.key.weight.qweight", "scales": "encoder.layer.3.attention.self.key.weight.scales", "biases": "encoder.layer.3.attention.self.key.weight.biases" }, { "name": "encoder.layer.3.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.3.attention.self.query.weight.qweight", "scales": "encoder.layer.3.attention.self.query.weight.scales", "biases": "encoder.layer.3.attention.self.query.weight.biases" }, { "name": "encoder.layer.3.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.3.attention.self.value.weight.qweight", "scales": "encoder.layer.3.attention.self.value.weight.scales", "biases": "encoder.layer.3.attention.self.value.weight.biases" }, { "name": "encoder.layer.3.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.3.intermediate.dense.weight.qweight", "scales": "encoder.layer.3.intermediate.dense.weight.scales", "biases": "encoder.layer.3.intermediate.dense.weight.biases" }, { "name": "encoder.layer.3.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.3.output.dense.weight.qweight", "scales": "encoder.layer.3.output.dense.weight.scales", "biases": "encoder.layer.3.output.dense.weight.biases" }, { "name": "encoder.layer.4.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.4.attention.output.dense.weight.qweight", "scales": "encoder.layer.4.attention.output.dense.weight.scales", "biases": "encoder.layer.4.attention.output.dense.weight.biases" }, { "name": "encoder.layer.4.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.4.attention.self.key.weight.qweight", "scales": "encoder.layer.4.attention.self.key.weight.scales", "biases": "encoder.layer.4.attention.self.key.weight.biases" }, { "name": "encoder.layer.4.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.4.attention.self.query.weight.qweight", "scales": "encoder.layer.4.attention.self.query.weight.scales", "biases": "encoder.layer.4.attention.self.query.weight.biases" }, { "name": "encoder.layer.4.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.4.attention.self.value.weight.qweight", "scales": "encoder.layer.4.attention.self.value.weight.scales", "biases": "encoder.layer.4.attention.self.value.weight.biases" }, { "name": "encoder.layer.4.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.4.intermediate.dense.weight.qweight", "scales": "encoder.layer.4.intermediate.dense.weight.scales", "biases": "encoder.layer.4.intermediate.dense.weight.biases" }, { "name": "encoder.layer.4.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.4.output.dense.weight.qweight", "scales": "encoder.layer.4.output.dense.weight.scales", "biases": "encoder.layer.4.output.dense.weight.biases" }, { "name": "encoder.layer.5.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.5.attention.output.dense.weight.qweight", "scales": "encoder.layer.5.attention.output.dense.weight.scales", "biases": "encoder.layer.5.attention.output.dense.weight.biases" }, { "name": "encoder.layer.5.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.5.attention.self.key.weight.qweight", "scales": "encoder.layer.5.attention.self.key.weight.scales", "biases": "encoder.layer.5.attention.self.key.weight.biases" }, { "name": "encoder.layer.5.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.5.attention.self.query.weight.qweight", "scales": "encoder.layer.5.attention.self.query.weight.scales", "biases": "encoder.layer.5.attention.self.query.weight.biases" }, { "name": "encoder.layer.5.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.5.attention.self.value.weight.qweight", "scales": "encoder.layer.5.attention.self.value.weight.scales", "biases": "encoder.layer.5.attention.self.value.weight.biases" }, { "name": "encoder.layer.5.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.5.intermediate.dense.weight.qweight", "scales": "encoder.layer.5.intermediate.dense.weight.scales", "biases": "encoder.layer.5.intermediate.dense.weight.biases" }, { "name": "encoder.layer.5.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.5.output.dense.weight.qweight", "scales": "encoder.layer.5.output.dense.weight.scales", "biases": "encoder.layer.5.output.dense.weight.biases" }, { "name": "encoder.layer.6.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.6.attention.output.dense.weight.qweight", "scales": "encoder.layer.6.attention.output.dense.weight.scales", "biases": "encoder.layer.6.attention.output.dense.weight.biases" }, { "name": "encoder.layer.6.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.6.attention.self.key.weight.qweight", "scales": "encoder.layer.6.attention.self.key.weight.scales", "biases": "encoder.layer.6.attention.self.key.weight.biases" }, { "name": "encoder.layer.6.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.6.attention.self.query.weight.qweight", "scales": "encoder.layer.6.attention.self.query.weight.scales", "biases": "encoder.layer.6.attention.self.query.weight.biases" }, { "name": "encoder.layer.6.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.6.attention.self.value.weight.qweight", "scales": "encoder.layer.6.attention.self.value.weight.scales", "biases": "encoder.layer.6.attention.self.value.weight.biases" }, { "name": "encoder.layer.6.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.6.intermediate.dense.weight.qweight", "scales": "encoder.layer.6.intermediate.dense.weight.scales", "biases": "encoder.layer.6.intermediate.dense.weight.biases" }, { "name": "encoder.layer.6.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.6.output.dense.weight.qweight", "scales": "encoder.layer.6.output.dense.weight.scales", "biases": "encoder.layer.6.output.dense.weight.biases" }, { "name": "encoder.layer.7.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.7.attention.output.dense.weight.qweight", "scales": "encoder.layer.7.attention.output.dense.weight.scales", "biases": "encoder.layer.7.attention.output.dense.weight.biases" }, { "name": "encoder.layer.7.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.7.attention.self.key.weight.qweight", "scales": "encoder.layer.7.attention.self.key.weight.scales", "biases": "encoder.layer.7.attention.self.key.weight.biases" }, { "name": "encoder.layer.7.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.7.attention.self.query.weight.qweight", "scales": "encoder.layer.7.attention.self.query.weight.scales", "biases": "encoder.layer.7.attention.self.query.weight.biases" }, { "name": "encoder.layer.7.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.7.attention.self.value.weight.qweight", "scales": "encoder.layer.7.attention.self.value.weight.scales", "biases": "encoder.layer.7.attention.self.value.weight.biases" }, { "name": "encoder.layer.7.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.7.intermediate.dense.weight.qweight", "scales": "encoder.layer.7.intermediate.dense.weight.scales", "biases": "encoder.layer.7.intermediate.dense.weight.biases" }, { "name": "encoder.layer.7.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.7.output.dense.weight.qweight", "scales": "encoder.layer.7.output.dense.weight.scales", "biases": "encoder.layer.7.output.dense.weight.biases" }, { "name": "encoder.layer.8.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.8.attention.output.dense.weight.qweight", "scales": "encoder.layer.8.attention.output.dense.weight.scales", "biases": "encoder.layer.8.attention.output.dense.weight.biases" }, { "name": "encoder.layer.8.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.8.attention.self.key.weight.qweight", "scales": "encoder.layer.8.attention.self.key.weight.scales", "biases": "encoder.layer.8.attention.self.key.weight.biases" }, { "name": "encoder.layer.8.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.8.attention.self.query.weight.qweight", "scales": "encoder.layer.8.attention.self.query.weight.scales", "biases": "encoder.layer.8.attention.self.query.weight.biases" }, { "name": "encoder.layer.8.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.8.attention.self.value.weight.qweight", "scales": "encoder.layer.8.attention.self.value.weight.scales", "biases": "encoder.layer.8.attention.self.value.weight.biases" }, { "name": "encoder.layer.8.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.8.intermediate.dense.weight.qweight", "scales": "encoder.layer.8.intermediate.dense.weight.scales", "biases": "encoder.layer.8.intermediate.dense.weight.biases" }, { "name": "encoder.layer.8.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.8.output.dense.weight.qweight", "scales": "encoder.layer.8.output.dense.weight.scales", "biases": "encoder.layer.8.output.dense.weight.biases" }, { "name": "encoder.layer.9.attention.output.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.9.attention.output.dense.weight.qweight", "scales": "encoder.layer.9.attention.output.dense.weight.scales", "biases": "encoder.layer.9.attention.output.dense.weight.biases" }, { "name": "encoder.layer.9.attention.self.key.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.9.attention.self.key.weight.qweight", "scales": "encoder.layer.9.attention.self.key.weight.scales", "biases": "encoder.layer.9.attention.self.key.weight.biases" }, { "name": "encoder.layer.9.attention.self.query.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.9.attention.self.query.weight.qweight", "scales": "encoder.layer.9.attention.self.query.weight.scales", "biases": "encoder.layer.9.attention.self.query.weight.biases" }, { "name": "encoder.layer.9.attention.self.value.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "encoder.layer.9.attention.self.value.weight.qweight", "scales": "encoder.layer.9.attention.self.value.weight.scales", "biases": "encoder.layer.9.attention.self.value.weight.biases" }, { "name": "encoder.layer.9.intermediate.dense.weight", "shape": [ 4096, 1024 ], "dtype": "float16", "qweight": "encoder.layer.9.intermediate.dense.weight.qweight", "scales": "encoder.layer.9.intermediate.dense.weight.scales", "biases": "encoder.layer.9.intermediate.dense.weight.biases" }, { "name": "encoder.layer.9.output.dense.weight", "shape": [ 1024, 4096 ], "dtype": "float16", "qweight": "encoder.layer.9.output.dense.weight.qweight", "scales": "encoder.layer.9.output.dense.weight.scales", "biases": "encoder.layer.9.output.dense.weight.biases" }, { "name": "pooler.dense.weight", "shape": [ 1024, 1024 ], "dtype": "float16", "qweight": "pooler.dense.weight.qweight", "scales": "pooler.dense.weight.scales", "biases": "pooler.dense.weight.biases" } ], "kept": [ { "name": "embeddings.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "embeddings.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "embeddings.position_ids", "shape": [ 1, 514 ], "dtype": "float16" }, { "name": "encoder.layer.0.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.0.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.0.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.1.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.1.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.10.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.10.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.11.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.11.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.12.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.12.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.13.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.13.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.14.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.14.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.15.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.15.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.16.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.16.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.17.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.17.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.18.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.18.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.19.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.19.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.2.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.2.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.20.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.20.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.21.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.21.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.22.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.22.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.23.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.23.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.3.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.3.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.4.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.4.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.5.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.5.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.6.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.6.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.7.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.7.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.8.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.8.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.attention.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.attention.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.attention.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.attention.self.key.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.attention.self.query.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.attention.self.value.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.intermediate.dense.bias", "shape": [ 4096 ], "dtype": "float16" }, { "name": "encoder.layer.9.output.LayerNorm.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.output.LayerNorm.weight", "shape": [ 1024 ], "dtype": "float16" }, { "name": "encoder.layer.9.output.dense.bias", "shape": [ 1024 ], "dtype": "float16" }, { "name": "pooler.dense.bias", "shape": [ 1024 ], "dtype": "float16" } ] }