{ "vocab_size_multiple": 8, "n_sample": 0, "src_vocab": "en.eole.vocab", "tgt_vocab_size": 32000, "save_data": "data", "tensorboard": true, "tensorboard_log_dir": "tensorboard", "overwrite": true, "transforms": [ "sentencepiece", "filtertoolong" ], "report_every": 100, "tensorboard_log_dir_dated": "tensorboard/Nov-18_22-06-12", "seed": 1234, "valid_metrics": [ "BLEU" ], "tgt_vocab": "is.eole.vocab", "share_vocab": false, "src_vocab_size": 32000, "training": { "warmup_steps": 5000, "attention_dropout": [ 0.1 ], "prefetch_factor": 32, "average_decay": 0.0001, "accum_count": [ 20 ], "gpu_ranks": [ 0 ], "dropout_steps": [ 0 ], "model_path": "quickmt-en-is-eole-model", "accum_steps": [ 0 ], "optim": "adamw", "batch_size": 6000, "keep_checkpoint": 4, "save_checkpoint_steps": 5000, "param_init_method": "xavier_uniform", "batch_size_multiple": 8, "world_size": 1, "compute_dtype": "torch.float16", "normalization": "tokens", "decay_method": "noam", "bucket_size": 128000, "learning_rate": 3.0, "valid_steps": 5000, "max_grad_norm": 0.0, "train_steps": 100000, "adam_beta2": 0.998, "num_workers": 0, "dropout": [ 0.1 ], "valid_batch_size": 2048, "batch_type": "tokens", "label_smoothing": 0.1 }, "data": { "corpus_1": { "path_src": "train_cefiltered5.en", "path_tgt": "train_cefiltered5.is", "transforms": [ "sentencepiece", "filtertoolong" ], "path_align": null }, "valid": { "path_src": "valid.en", "path_tgt": "valid.is", "transforms": [ "sentencepiece", "filtertoolong" ], "path_align": null } }, "transforms_configs": { "sentencepiece": { "tgt_subword_model": "${MODEL_PATH}/is.spm.model", "src_subword_model": "${MODEL_PATH}/en.spm.model" }, "filtertoolong": { "tgt_seq_length": 256, "src_seq_length": 256 } }, "model": { "hidden_size": 1024, "share_decoder_embeddings": true, "architecture": "transformer", "position_encoding_type": "SinusoidalInterleaved", "heads": 8, "share_embeddings": false, "transformer_ff": 4096, "encoder": { "hidden_size": 1024, "position_encoding_type": "SinusoidalInterleaved", "heads": 8, "layers": 8, "encoder_type": "transformer", "src_word_vec_size": 1024, "transformer_ff": 4096, "n_positions": null }, "embeddings": { "src_word_vec_size": 1024, "word_vec_size": 1024, "position_encoding_type": "SinusoidalInterleaved", "tgt_word_vec_size": 1024 }, "decoder": { "hidden_size": 1024, "position_encoding_type": "SinusoidalInterleaved", "heads": 8, "layers": 2, "n_positions": null, "decoder_type": "transformer", "transformer_ff": 4096, "tgt_word_vec_size": 1024 } } }