decompattn_graph.json

{
  "dataset_reader": {
    "type": "scitail_graph",
    "token_indexers": {
      "tokens": {
        "type": "single_id",
        "lowercase_tokens": true
      }
    },
    "tokenizer": {
      "end_tokens": ["@@NULL@@"]
    },
    "entities_tokenizer": {
      "word_splitter": {
        "type": "entities_splitter"
      }
    }
  },
  // relative to project root
  "train_data_path":
    "../CSNLI/data/nli_datasets/scitail/conceptnet_entities/parsed/scitail_1.0_train_conceptnet_entities.tsv",
  "validation_data_path":
    "../CSNLI/data/nli_datasets/scitail/conceptnet_entities/parsed/scitail_1.0_dev_conceptnet_entities.tsv",
  "model": {
    "type": "graph_and_text",
    "text_model": {
      "type": "decompattn",
      "text_field_embedder": {
        "tokens": {
          "type": "embedding",
          "projection_dim": 200,
          "pretrained_file": "./data/embeddings/glove/glove.840B.300d.txt.gz",
          "embedding_dim": 300,
          "trainable": false
        }
      },
      "attend_feedforward": {
        "input_dim": 200,
        "num_layers": 2,
        "hidden_dims": 200,
        "activations": "relu",
        "dropout": 0.2
      },
      "similarity_function": { "type": "dot_product" },
      "compare_feedforward": {
        "input_dim": 400,
        "num_layers": 2,
        "hidden_dims": 200,
        "activations": "relu",
        "dropout": 0.2
      },
      "aggregate_feedforward": {
        "input_dim": 400,
        "num_layers": 2,
        "hidden_dims": [200, 2],
        "activations": ["relu", "linear"],
        "dropout": [0.2, 0.0]
      },
      "initializer": [
        [".*linear_layers.*weight", { "type": "xavier_normal" }],
        [
          ".*token_embedder_tokens._projection.*weight",
          { "type": "xavier_normal" }
        ]
      ]
    },
    "graph_model": {
      "type": "simple_graph",
      "entities_embedder": {
        "tokens": {
          "type": "embedding",
          "projection_dim": 200,
          "pretrained_file": "./data/embeddings/conceptnet/transh.txt.gz",
          "embedding_dim": 300,
          "trainable": false
        }
      },
      "inter_attention": {
        "type": "dot_product"
      },
      "project_feedforward": {
        "input_dim": 800,
        "num_layers": 2,
        "hidden_dims": 200,
        "activations": "relu",
        "dropout": 0.2
      },
      "aggregate_feedforward": {
        "input_dim": 800,
        "num_layers": 2,
        "hidden_dims": [200, 2],
        "activations": ["relu", "linear"],
        "dropout": [0.2, 0.0]
      },
      "initializer": [
        [".*linear_layers.*weight", { "type": "xavier_normal" }],
        [
          ".*token_embedder_tokens._projection.*weight",
          { "type": "xavier_normal" }
        ]
      ]
    },
    "classify_feed_forward": {
      "input_dim": 1200,
      "num_layers": 2,
      "hidden_dims": [600, 2],
      "activations": ["relu", "linear"],
      "dropout": [0.2, 0.0]
    }
  },
  "iterator": {
    "type": "bucket",
    "sorting_keys": [
      ["premise", "num_tokens"],
      ["hypothesis", "num_tokens"],
      ["premise_entities", "num_tokens"],
      ["hypothesis_entities", "num_tokens"]
    ],
    "batch_size": 64
  },

  "trainer": {
    "num_epochs": 140,
    "patience": 20,
    "cuda_device": -1,
    "grad_clipping": 5.0,
    "validation_metric": "+accuracy",
    "optimizer": {
      "type": "adagrad"
    }
  }
}