_data/architectures/roberta-masked-lm.json

{
    "model_type": "roberta",
    "architectures": [
        "RobertaForMaskedLM"
    ],
    "pre_weights": [
        {
            "name": "roberta.embeddings.position_embeddings.weight"
        },
        {
            "name": "roberta.embeddings.word_embeddings.weight",
            "is_embed": true
        },
        {
            "name": "roberta.embeddings.token_type_embeddings.weight"
        },
        {
            "name": "roberta.embeddings.LayerNorm.weight"
        },
        {
            "name": "roberta.embeddings.LayerNorm.bias"
        },
        {
            "name": "roberta.embeddings.position_ids",
            "optional": true,
            "force_dtype": "int64"
        }
    ],
    "post_weights": [
        {
            "name": "lm_head.bias"
        },
        {
            "name": "lm_head.dense.weight"
        },
        {
            "name": "lm_head.dense.bias"
        },
        {
            "name": "lm_head.layer_norm.weight"
        },
        {
            "name": "lm_head.layer_norm.bias"
        },
        {
            "name": "lm_head.decoder.weight",
            "is_embed": true,
            "optional": true,
            "tied_names": [
                "roberta.embeddings.word_embeddings.weight"
            ]
        }
    ],
    "num_layers_config_key": "num_hidden_layers",
    "layer_templates": {
        "weights": [
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.output.dense.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.output.dense.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.output.LayerNorm.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.output.LayerNorm.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.self.query.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.self.query.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.self.key.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.self.key.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.self.value.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.attention.self.value.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.intermediate.dense.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.intermediate.dense.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.output.dense.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.output.dense.bias"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.output.LayerNorm.weight"
            },
            {
                "name": "roberta.encoder.layer.${layer_index}.output.LayerNorm.bias"
            }
        ]
    }
}