{ "tensorboard_log_dir": "tensorboard", "tgt_vocab": "ja.eole.vocab", "share_vocab": false, "vocab_size_multiple": 8, "n_sample": 0, "src_vocab": "en.eole.vocab", "transforms": [ "sentencepiece", "filtertoolong" ], "report_every": 100, "tgt_vocab_size": 20000, "tensorboard_log_dir_dated": "tensorboard/Apr-17_21-24-27", "save_data": "data", "src_vocab_size": 20000, "overwrite": true, "tensorboard": true, "valid_metrics": [ "BLEU" ], "seed": 1234, "training": { "attention_dropout": [ 0.1 ], "dropout_steps": [ 0 ], "normalization": "tokens", "max_grad_norm": 0.0, "decay_method": "noam", "bucket_size": 128000, "average_decay": 0.0, "prefetch_factor": 32, "learning_rate": 2.0, "valid_batch_size": 4096, "optim": "adamw", "warmup_steps": 4000, "adam_beta2": 0.998, "batch_type": "tokens", "num_workers": 0, "save_checkpoint_steps": 5000, "param_init_method": "xavier_uniform", "accum_steps": [ 0 ], "train_steps": 100000, "gpu_ranks": [ 0 ], "model_path": "quickmt-en-ja-eole-model", "dropout": [ 0.1 ], "compute_dtype": "torch.float16", "keep_checkpoint": 4, "batch_size": 8000, "valid_steps": 5000, "world_size": 1, "accum_count": [ 10 ], "batch_size_multiple": 8, "label_smoothing": 0.1 }, "data": { "corpus_1": { "path_src": "en.txt", "path_tgt": "ja.txt", "transforms": [ "sentencepiece", "filtertoolong" ], "path_align": null }, "valid": { "path_src": "valid.en", "path_tgt": "valid.ja", "transforms": [ "sentencepiece", "filtertoolong" ], "path_align": null } }, "transforms_configs": { "sentencepiece": { "src_subword_model": "${MODEL_PATH}/en.spm.model", "tgt_subword_model": "${MODEL_PATH}/ja.spm.model" }, "filtertoolong": { "tgt_seq_length": 256, "src_seq_length": 256 } }, "model": { "hidden_size": 1024, "position_encoding_type": "SinusoidalInterleaved", "architecture": "transformer", "heads": 8, "transformer_ff": 4096, "share_embeddings": false, "share_decoder_embeddings": false, "encoder": { "n_positions": null, "position_encoding_type": "SinusoidalInterleaved", "heads": 8, "encoder_type": "transformer", "layers": 8, "transformer_ff": 4096, "hidden_size": 1024, "src_word_vec_size": 1024 }, "embeddings": { "tgt_word_vec_size": 1024, "word_vec_size": 1024, "src_word_vec_size": 1024, "position_encoding_type": "SinusoidalInterleaved" }, "decoder": { "n_positions": null, "position_encoding_type": "SinusoidalInterleaved", "heads": 8, "layers": 2, "tgt_word_vec_size": 1024, "decoder_type": "transformer", "transformer_ff": 4096, "hidden_size": 1024 } } }