quickmt/quickmt-en-fa
287
1{2 "seed": 1234,3 "save_data": "data",4 "tgt_vocab": "fa.eole.vocab",5 "transforms": [6 "sentencepiece",7 "filtertoolong"8 ],9 "n_sample": 0,10 "vocab_size_multiple": 8,11 "tensorboard_log_dir": "tensorboard",12 "src_vocab_size": 20000,13 "tgt_vocab_size": 20000,14 "report_every": 100,15 "overwrite": true,16 "share_vocab": false,17 "tensorboard": true,18 "src_vocab": "en.eole.vocab",19 "tensorboard_log_dir_dated": "tensorboard/Jul-06_14-51-35",20 "valid_metrics": [21 "BLEU"22 ],23 "training": {24 "valid_steps": 5000,25 "warmup_steps": 4000,26 "max_grad_norm": 0.0,27 "average_decay": 0.0001,28 "normalization": "tokens",29 "num_workers": 0,30 "prefetch_factor": 32,31 "dropout": [32 0.133 ],34 "decay_method": "noam",35 "batch_type": "tokens",36 "train_steps": 100000,37 "batch_size": 8000,38 "save_checkpoint_steps": 5000,39 "accum_steps": [40 041 ],42 "compute_dtype": "torch.float16",43 "world_size": 1,44 "model_path": "quickmt-en-fa-eole-model",45 "optim": "adamw",46 "param_init_method": "xavier_uniform",47 "accum_count": [48 1049 ],50 "adam_beta2": 0.998,51 "gpu_ranks": [52 053 ],54 "label_smoothing": 0.1,55 "dropout_steps": [56 057 ],58 "bucket_size": 128000,59 "batch_size_multiple": 8,60 "attention_dropout": [61 0.162 ],63 "learning_rate": 2.0,64 "keep_checkpoint": 5,65 "valid_batch_size": 409666 },67 "data": {68 "corpus_1": {69 "path_align": null,70 "path_src": "train.en",71 "path_tgt": "train.fa",72 "transforms": [73 "sentencepiece",74 "filtertoolong"75 ]76 },77 "valid": {78 "path_align": null,79 "path_src": "dev.en",80 "path_tgt": "dev.fa",81 "transforms": [82 "sentencepiece",83 "filtertoolong"84 ]85 }86 },87 "transforms_configs": {88 "sentencepiece": {89 "src_subword_model": "${MODEL_PATH}/en.spm.model",90 "tgt_subword_model": "${MODEL_PATH}/fa.spm.model"91 },92 "filtertoolong": {93 "src_seq_length": 256,94 "tgt_seq_length": 25695 }96 },97 "model": {98 "heads": 8,99 "position_encoding_type": "SinusoidalInterleaved",100 "hidden_size": 1024,101 "share_embeddings": false,102 "transformer_ff": 4096,103 "share_decoder_embeddings": false,104 "architecture": "transformer",105 "decoder": {106 "heads": 8,107 "tgt_word_vec_size": 1024,108 "position_encoding_type": "SinusoidalInterleaved",109 "hidden_size": 1024,110 "n_positions": null,111 "transformer_ff": 4096,112 "decoder_type": "transformer",113 "layers": 2114 },115 "encoder": {116 "encoder_type": "transformer",117 "heads": 8,118 "position_encoding_type": "SinusoidalInterleaved",119 "hidden_size": 1024,120 "n_positions": null,121 "transformer_ff": 4096,122 "layers": 8,123 "src_word_vec_size": 1024124 },125 "embeddings": {126 "tgt_word_vec_size": 1024,127 "src_word_vec_size": 1024,128 "word_vec_size": 1024,129 "position_encoding_type": "SinusoidalInterleaved"130 }131 }132}