{ "seed": 1234, "save_data": "data", "tgt_vocab": "fa.eole.vocab", "transforms": [ "sentencepiece", "filtertoolong" ], "n_sample": 0, "vocab_size_multiple": 8, "tensorboard_log_dir": "tensorboard", "src_vocab_size": 20000, "tgt_vocab_size": 20000, "report_every": 100, "overwrite": true, "share_vocab": false, "tensorboard": true, "src_vocab": "en.eole.vocab", "tensorboard_log_dir_dated": "tensorboard/Jul-06_14-51-35", "valid_metrics": [ "BLEU" ], "training": { "valid_steps": 5000, "warmup_steps": 4000, "max_grad_norm": 0.0, "average_decay": 0.0001, "normalization": "tokens", "num_workers": 0, "prefetch_factor": 32, "dropout": [ 0.1 ], "decay_method": "noam", "batch_type": "tokens", "train_steps": 100000, "batch_size": 8000, "save_checkpoint_steps": 5000, "accum_steps": [ 0 ], "compute_dtype": "torch.float16", "world_size": 1, "model_path": "quickmt-en-fa-eole-model", "optim": "adamw", "param_init_method": "xavier_uniform", "accum_count": [ 10 ], "adam_beta2": 0.998, "gpu_ranks": [ 0 ], "label_smoothing": 0.1, "dropout_steps": [ 0 ], "bucket_size": 128000, "batch_size_multiple": 8, "attention_dropout": [ 0.1 ], "learning_rate": 2.0, "keep_checkpoint": 5, "valid_batch_size": 4096 }, "data": { "corpus_1": { "path_align": null, "path_src": "train.en", "path_tgt": "train.fa", "transforms": [ "sentencepiece", "filtertoolong" ] }, "valid": { "path_align": null, "path_src": "dev.en", "path_tgt": "dev.fa", "transforms": [ "sentencepiece", "filtertoolong" ] } }, "transforms_configs": { "sentencepiece": { "src_subword_model": "${MODEL_PATH}/en.spm.model", "tgt_subword_model": "${MODEL_PATH}/fa.spm.model" }, "filtertoolong": { "src_seq_length": 256, "tgt_seq_length": 256 } }, "model": { "heads": 8, "position_encoding_type": "SinusoidalInterleaved", "hidden_size": 1024, "share_embeddings": false, "transformer_ff": 4096, "share_decoder_embeddings": false, "architecture": "transformer", "decoder": { "heads": 8, "tgt_word_vec_size": 1024, "position_encoding_type": "SinusoidalInterleaved", "hidden_size": 1024, "n_positions": null, "transformer_ff": 4096, "decoder_type": "transformer", "layers": 2 }, "encoder": { "encoder_type": "transformer", "heads": 8, "position_encoding_type": "SinusoidalInterleaved", "hidden_size": 1024, "n_positions": null, "transformer_ff": 4096, "layers": 8, "src_word_vec_size": 1024 }, "embeddings": { "tgt_word_vec_size": 1024, "src_word_vec_size": 1024, "word_vec_size": 1024, "position_encoding_type": "SinusoidalInterleaved" } } }