TwoTowerClassificationModel.Config

Component: TwoTowerClassificationModel

class TwoTowerClassificationModel.Config

Bases: BaseModel.Config

All Attributes (including base classes)

inputs: InputConfig = InputConfig()

right_encoder: RoBERTaEncoderBase.Config = RoBERTaEncoder.Config()

left_encoder: RoBERTaEncoderBase.Config = RoBERTaEncoder.Config()

decoder: MLPDecoderTwoTower.Config = MLPDecoderTwoTower.Config()

output_layer: ClassificationOutputLayer.Config = ClassificationOutputLayer.Config()

Default JSON

{
    "inputs": {
        "right_tokens": {
            "RoBERTaTensorizer": {
                "is_input": true,
                "columns": [
                    "text"
                ],
                "tokenizer": {
                    "GPT2BPETokenizer": {
                        "bpe_encoder_path": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/encoder.json",
                        "bpe_vocab_path": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/vocab.bpe",
                        "lowercase": false
                    }
                },
                "base_tokenizer": null,
                "vocab_file": "gpt2_bpe_dict",
                "max_seq_len": 256,
                "add_selfie_token": false
            }
        },
        "left_tokens": {
            "RoBERTaTensorizer": {
                "is_input": true,
                "columns": [
                    "text"
                ],
                "tokenizer": {
                    "GPT2BPETokenizer": {
                        "bpe_encoder_path": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/encoder.json",
                        "bpe_vocab_path": "manifold://pytext_training/tree/static/vocabs/bpe/gpt2/vocab.bpe",
                        "lowercase": false
                    }
                },
                "base_tokenizer": null,
                "vocab_file": "gpt2_bpe_dict",
                "max_seq_len": 256,
                "add_selfie_token": false
            }
        },
        "right_dense": null,
        "left_dense": null,
        "labels": {
            "LabelTensorizer": {
                "is_input": false,
                "column": "label",
                "allow_unknown": false,
                "pad_in_vocab": false,
                "label_vocab": null,
                "label_vocab_file": null,
                "add_labels": null
            }
        }
    },
    "right_encoder": {
        "RoBERTaEncoder": {
            "load_path": null,
            "save_path": null,
            "freeze": false,
            "shared_module_key": null,
            "output_dropout": 0.4,
            "embedding_dim": 768,
            "pooling": "cls_token",
            "export": false,
            "projection_dim": 0,
            "normalize_output_rep": false,
            "vocab_size": 50265,
            "num_encoder_layers": 12,
            "num_attention_heads": 12,
            "model_path": "manifold://pytext_training/tree/static/models/roberta_base_torch.pt",
            "is_finetuned": false,
            "max_seq_len": 514,
            "use_bias_finetuning": false,
            "use_linformer_encoder": false,
            "linformer_compressed_ratio": 4,
            "linformer_quantize": false,
            "export_encoder": false,
            "variable_size_embedding": true,
            "use_selfie_encoder": false,
            "transformer_layer_to_keep": null,
            "attention_heads_to_keep_per_layer": null
        }
    },
    "left_encoder": {
        "RoBERTaEncoder": {
            "load_path": null,
            "save_path": null,
            "freeze": false,
            "shared_module_key": null,
            "output_dropout": 0.4,
            "embedding_dim": 768,
            "pooling": "cls_token",
            "export": false,
            "projection_dim": 0,
            "normalize_output_rep": false,
            "vocab_size": 50265,
            "num_encoder_layers": 12,
            "num_attention_heads": 12,
            "model_path": "manifold://pytext_training/tree/static/models/roberta_base_torch.pt",
            "is_finetuned": false,
            "max_seq_len": 514,
            "use_bias_finetuning": false,
            "use_linformer_encoder": false,
            "linformer_compressed_ratio": 4,
            "linformer_quantize": false,
            "export_encoder": false,
            "variable_size_embedding": true,
            "use_selfie_encoder": false,
            "transformer_layer_to_keep": null,
            "attention_heads_to_keep_per_layer": null
        }
    },
    "decoder": {
        "load_path": null,
        "save_path": null,
        "freeze": false,
        "shared_module_key": null,
        "right_hidden_dims": [],
        "left_hidden_dims": [],
        "hidden_dims": [],
        "layer_norm": false,
        "dropout": 0.0
    },
    "output_layer": {
        "load_path": null,
        "save_path": null,
        "freeze": false,
        "shared_module_key": null,
        "loss": {
            "CrossEntropyLoss": {}
        },
        "label_weights": null
    }
}