allen_conf_vocab.json

{
  "dataset_reader": {
    "type": "mention_categories",
    "category_mapping_file": "./data/category_mapping.json",
    "tokenizer": {
      "type": "word",
      "word_splitter": {
        "type": "fast-splitter"
      }
    },
    "token_indexers": {
      "tokens": {
        "type": "fasttext-id",
        "model_path": "./data/fasttext_embedding.model",
        "model_params_path": "./data/fasttext_embedding.model.params"
      }
    }
  },
  "train_data_path": "./data/train_data.tsv",
  "validation_data_path": "./data/valid_data.tsv",
  "evaluate_on_test": true,
  "model": {
    "type": "category_predictor",
    "text_embedder": {
      "type": "basic",
      "token_embedders": {
        "type": "basic",
        "allow_unmatched_keys": true,
        "embedder_to_indexer_map": {
          "tokens-ngram": [
            "tokens-ngram",
            "tokens-ngram-lengths",
            "tokens-ngram-mask"
          ],
          "tokens": [
            "tokens"
          ]
        },
        "token_embedders": {
          "tokens-ngram": {
            "type": "fasttext-embedder",
            "model_path": "./data/fasttext_embedding.model",
            "trainable": false
          },
          "tokens": {
            "type": "embedding",
            "embedding_dim": 10,
            "trainable": true
          }
        }
      }
    },
    "encoder": {
      "type": "lstm",
      "input_size": 310,
      "hidden_size": 50,
      "num_layers": 1,
      "dropout": 0.5,
      "bidirectional": true
    },
    "seq_combiner": {
      "type": "attention-combiner",
      "num_layers": 2,
      "input_state_dim": 100,
      "encoded_seq_dim": 100,
      "output_dim": 50,
      "dropout": 0.2,
      "attention": {
        "type": "linear",
        "tensor_1_dim": 100,
        "tensor_2_dim": 100
      }
    }
  },
  "iterator": {
    "type": "numpy-multiprocessing",
    "num_workers": 2,
    "base_iterator": {
      "type": "basic",
      "batch_size": 32
    }
  },
  "trainer": {
    "shuffle": false,
    "optimizer": {
      "type": "adam",
      "lr": 0.001
    },
    "should_log_learning_rate": true,
    "num_epochs": 50,
    "learning_rate_scheduler": {
      "type": "reduce_on_plateau",
      "factor": 0.5,
      "patience": 5,
      "verbose": true
    }
  },
  "vocabulary": {
    "directory_path": null,
    "min_count": {
      "tokens": 3
    },
    "max_vocab_size": 200000
  }
}