config.json

{
  "_name_or_path": "trained_models/t5_qanom-joint-23.03.22",
  "append_verb_form": true,
  "architectures": [
    "T5ForConditionalGeneration"
  ],
  "d_ff": 2048,
  "d_kv": 64,
  "d_model": 512,
  "debug_mode": false,
  "decoder_start_token_id": 0,
  "dense_act_fn": "relu",
  "description": "optimal joint config from sweep, mainly for qanom",
  "dir_switch": "joint_optimal",
  "do_eval_on": "validation",
  "dropout_rate": 0.1,
  "eos_token_id": 1,
  "eval_steps": 500,
  "evaluation_strategy": "steps",
  "feed_forward_proj": "relu",
  "fp16": true,
  "gradient_accumulation_steps": 14,
  "initializer_factor": 1.0,
  "is_encoder_decoder": true,
  "is_gated_act": false,
  "layer_norm_epsilon": 1e-06,
  "learning_rate": 0.001,
  "load_best_model_at_end": true,
  "logging_steps": 500,
  "logging_strategy": "steps",
  "metric_for_best_model": "eval_loss",
  "model_type": "t5",
  "n_positions": 512,
  "num_beams": 5,
  "num_decoder_layers": 6,
  "num_heads": 8,
  "num_layers": 6,
  "output_past": true,
  "overwrite_output_dir": true,
  "pad_token_id": 0,
  "per_device_eval_batch_size": 12,
  "per_device_train_batch_size": 12,
  "predicate_marker_type": "generic",
  "predict_with_generate": true,
  "preprocess_input_func": "input_predicate_marker",
  "preprocessing_kwargs": {
    "append_verb_form": true,
    "debug_mode": false,
    "description": "optimal joint config from sweep, mainly for qanom",
    "dir_switch": "joint_optimal",
    "do_eval_on": "validation",
    "dropout_rate": 0.1,
    "eval_steps": 500,
    "evaluation_strategy": "steps",
    "fp16": true,
    "gradient_accumulation_steps": 14,
    "learning_rate": 0.001,
    "load_best_model_at_end": true,
    "logging_steps": 500,
    "logging_strategy": "steps",
    "metric_for_best_model": "eval_loss",
    "model_type": "t5",
    "num_beams": 5,
    "overwrite_output_dir": true,
    "per_device_eval_batch_size": 12,
    "per_device_train_batch_size": 12,
    "predicate_marker_type": "generic",
    "predict_with_generate": true,
    "preprocess_input_func": "input_predicate_marker",
    "qanom_joint_factor": 14,
    "save_steps": 500,
    "save_strategy": "steps",
    "seed": 44,
    "source_prefix": "parse: ",
    "train_dataset": "joint_qanom",
    "train_epochs": 20,
    "use_bilateral_predicate_marker": true
  },
  "qanom_joint_factor": 14,
  "relative_attention_max_distance": 128,
  "relative_attention_num_buckets": 32,
  "save_steps": 500,
  "save_strategy": "steps",
  "seed": 44,
  "source_prefix": "parse: ",
  "task_specific_params": {
    "summarization": {
      "early_stopping": true,
      "length_penalty": 2.0,
      "max_length": 200,
      "min_length": 30,
      "no_repeat_ngram_size": 3,
      "num_beams": 4,
      "prefix": "summarize: "
    },
    "translation_en_to_de": {
      "early_stopping": true,
      "max_length": 300,
      "num_beams": 4,
      "prefix": "translate English to German: "
    },
    "translation_en_to_fr": {
      "early_stopping": true,
      "max_length": 300,
      "num_beams": 4,
      "prefix": "translate English to French: "
    },
    "translation_en_to_ro": {
      "early_stopping": true,
      "max_length": 300,
      "num_beams": 4,
      "prefix": "translate English to Romanian: "
    }
  },
  "torch_dtype": "float32",
  "train_dataset": "joint_qanom",
  "train_epochs": 20,
  "transformers_version": "4.26.1",
  "use_bilateral_predicate_marker": true,
  "use_cache": true,
  "vocab_size": 32101
}