{ "_name_or_path": "Python/ACROSS-m2o-eng-small", "architectures": [ "MT5ForConditionalGeneration" ], "consistency_loss_weight": 2.0, "consistency_temperature": 1.0, "contrastive": true, "contrastive_mini_batch_size": 4, "cross_alignment_loss_weight": 1.0, "d_ff": 1024, "d_kv": 64, "d_model": 512, "decoder_start_token_id": 0, "dropout_rate": 0.1, "eos_token_id": 1, "feature_pooling_method": "mean_pooling", "feed_forward_proj": "gated-gelu", "freeze_decoder": false, "freeze_mono_encoder": true, "initializer_factor": 1.0, "is_encoder_decoder": true, "layer_norm_epsilon": 1e-06, "length_penalty": 0.6, "max_length": 84, "model_type": "mt5", "num_beams": 4, "num_decoder_layers": 8, "num_heads": 6, "num_layers": 8, "pad_token_id": 0, "queue_size": 8192, "relative_attention_num_buckets": 32, "similarity_method": "cosine", "t5_path_or_url": "output/eng2eng_small_baseline_A5000", "task_specific_params": { "langid_map": { "amharic": [ 39, "\u2581" ], "arabic": [ 8, "\u2581" ], "azerbaijani": [ 11, "\u2581" ], "bengali": [ 46, "\u2581" ], "burmese": [ 37, "\u2581" ], "chinese_simplified": [ 44, "\u2581" ], "chinese_traditional": [ 48, "\u2581" ], "czech": [ 3, "\u2581" ], "dutch": [ 1, "\u2581" ], "english": [ 34, "\u2581" ], "french": [ 14, "\u2581" ], "german": [ 0, "\u2581" ], "gujarati": [ 31, "\u2581" ], "hausa": [ 47, "\u2581" ], "hindi": [ 25, "\u2581" ], "igbo": [ 13, "\u2581" ], "indonesian": [ 5, "\u2581" ], "italian": [ 2, "\u2581" ], "japanese": [ 41, "\u2581" ], "kirundi": [ 4, "\u2581" ], "korean": [ 33, "\u2581" ], "kyrgyz": [ 9, "\u2581" ], "marathi": [ 17, "\u2581" ], "nepali": [ 24, "\u2581" ], "oromo": [ 45, "\u2581" ], "pashto": [ 38, "\u2581" ], "persian": [ 27, "\u2581" ], "pidgin": [ 18, "\u2581" ], "portuguese": [ 43, "\u2581" ], "punjabi": [ 21, "\u2581" ], "russian": [ 40, "\u2581" ], "scottish_gaelic": [ 28, "\u2581" ], "serbian_cyrillic": [ 32, "\u2581" ], "serbian_latin": [ 15, "\u2581" ], "sinhala": [ 35, "\u2581" ], "somali": [ 23, "\u2581" ], "spanish": [ 7, "\u2581" ], "swahili": [ 22, "\u2581" ], "tamil": [ 36, "\u2581" ], "telugu": [ 26, "\u2581" ], "thai": [ 10, "\u2581" ], "tigrinya": [ 20, "\u2581" ], "turkish": [ 19, "\u2581" ], "ukrainian": [ 6, "\u2581" ], "urdu": [ 42, "\u2581" ], "uzbek": [ 12, "\u2581" ], "vietnamese": [ 16, "\u2581" ], "welsh": [ 30, "\u2581" ], "yoruba": [ 29, "\u2581" ] } }, "temperature": 0.1, "tie_word_embeddings": false, "tokenizer_class": "T5Tokenizer", "use_cache": true, "use_consistency_loss": true, "use_cross_alignment_loss": true, "use_queue_for_contrastive": true, "vocab_size": 250112 }