| !ModelConfig | |
| config_data: !DataConfig | |
| data_statistics: !DataStatistics | |
| average_len_target_per_bucket: | |
| - 6.011277314861252 | |
| - 10.39671010650862 | |
| - 18.504618481112797 | |
| - 26.041977712863865 | |
| - 32.48293857888395 | |
| - 41.60258525852575 | |
| - 51.3669121514993 | |
| - 59.76513060097686 | |
| - 68.25156398104271 | |
| - 75.67976141505562 | |
| - 83.09476309226905 | |
| - 21.1875 | |
| - 7.749999999999999 | |
| - 5.833333333333333 | |
| - 6.0 | |
| - 7.0 | |
| - 7.25 | |
| - 6.333333333333333 | |
| - 7.0 | |
| - 5.0 | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| - null | |
| buckets: | |
| - !!python/tuple | |
| - 8 | |
| - 8 | |
| - !!python/tuple | |
| - 16 | |
| - 16 | |
| - !!python/tuple | |
| - 24 | |
| - 24 | |
| - !!python/tuple | |
| - 32 | |
| - 32 | |
| - !!python/tuple | |
| - 40 | |
| - 40 | |
| - !!python/tuple | |
| - 48 | |
| - 48 | |
| - !!python/tuple | |
| - 56 | |
| - 56 | |
| - !!python/tuple | |
| - 64 | |
| - 64 | |
| - !!python/tuple | |
| - 72 | |
| - 72 | |
| - !!python/tuple | |
| - 80 | |
| - 80 | |
| - !!python/tuple | |
| - 88 | |
| - 88 | |
| - !!python/tuple | |
| - 96 | |
| - 96 | |
| - !!python/tuple | |
| - 104 | |
| - 104 | |
| - !!python/tuple | |
| - 112 | |
| - 112 | |
| - !!python/tuple | |
| - 120 | |
| - 120 | |
| - !!python/tuple | |
| - 128 | |
| - 128 | |
| - !!python/tuple | |
| - 136 | |
| - 136 | |
| - !!python/tuple | |
| - 144 | |
| - 144 | |
| - !!python/tuple | |
| - 152 | |
| - 152 | |
| - !!python/tuple | |
| - 160 | |
| - 160 | |
| - !!python/tuple | |
| - 168 | |
| - 168 | |
| - !!python/tuple | |
| - 176 | |
| - 176 | |
| - !!python/tuple | |
| - 184 | |
| - 184 | |
| - !!python/tuple | |
| - 192 | |
| - 192 | |
| - !!python/tuple | |
| - 200 | |
| - 200 | |
| - !!python/tuple | |
| - 208 | |
| - 208 | |
| - !!python/tuple | |
| - 216 | |
| - 216 | |
| - !!python/tuple | |
| - 224 | |
| - 224 | |
| - !!python/tuple | |
| - 232 | |
| - 232 | |
| - !!python/tuple | |
| - 240 | |
| - 240 | |
| - !!python/tuple | |
| - 248 | |
| - 248 | |
| - !!python/tuple | |
| - 256 | |
| - 256 | |
| - !!python/tuple | |
| - 264 | |
| - 264 | |
| - !!python/tuple | |
| - 272 | |
| - 272 | |
| - !!python/tuple | |
| - 280 | |
| - 280 | |
| - !!python/tuple | |
| - 288 | |
| - 288 | |
| - !!python/tuple | |
| - 296 | |
| - 296 | |
| - !!python/tuple | |
| - 304 | |
| - 304 | |
| - !!python/tuple | |
| - 312 | |
| - 312 | |
| - !!python/tuple | |
| - 320 | |
| - 320 | |
| - !!python/tuple | |
| - 328 | |
| - 328 | |
| - !!python/tuple | |
| - 336 | |
| - 336 | |
| - !!python/tuple | |
| - 344 | |
| - 344 | |
| - !!python/tuple | |
| - 352 | |
| - 352 | |
| - !!python/tuple | |
| - 360 | |
| - 360 | |
| - !!python/tuple | |
| - 368 | |
| - 368 | |
| - !!python/tuple | |
| - 376 | |
| - 376 | |
| - !!python/tuple | |
| - 384 | |
| - 384 | |
| - !!python/tuple | |
| - 392 | |
| - 392 | |
| - !!python/tuple | |
| - 400 | |
| - 400 | |
| - !!python/tuple | |
| - 408 | |
| - 408 | |
| - !!python/tuple | |
| - 416 | |
| - 416 | |
| - !!python/tuple | |
| - 424 | |
| - 424 | |
| - !!python/tuple | |
| - 432 | |
| - 432 | |
| - !!python/tuple | |
| - 440 | |
| - 440 | |
| - !!python/tuple | |
| - 448 | |
| - 448 | |
| - !!python/tuple | |
| - 456 | |
| - 456 | |
| - !!python/tuple | |
| - 464 | |
| - 464 | |
| - !!python/tuple | |
| - 472 | |
| - 472 | |
| - !!python/tuple | |
| - 480 | |
| - 480 | |
| - !!python/tuple | |
| - 488 | |
| - 488 | |
| - !!python/tuple | |
| - 496 | |
| - 496 | |
| - !!python/tuple | |
| - 504 | |
| - 504 | |
| - !!python/tuple | |
| - 512 | |
| - 512 | |
| - !!python/tuple | |
| - 513 | |
| - 513 | |
| length_ratio_mean: 1.4462363190719616 | |
| length_ratio_stats_per_bucket: | |
| - !!python/tuple | |
| - 1.13785261776882 | |
| - 0.30617192612461125 | |
| - !!python/tuple | |
| - 1.6532001104346803 | |
| - 0.6379238313847178 | |
| - !!python/tuple | |
| - 2.471066932954014 | |
| - 0.9542933578872858 | |
| - !!python/tuple | |
| - 2.9657919123449297 | |
| - 1.5793440289847793 | |
| - !!python/tuple | |
| - 3.096168086847911 | |
| - 2.441325976657657 | |
| - !!python/tuple | |
| - 3.268248874962792 | |
| - 2.2403195790511785 | |
| - !!python/tuple | |
| - 3.453956732311208 | |
| - 2.518686961607376 | |
| - !!python/tuple | |
| - 3.2456624469723585 | |
| - 2.2619507905455962 | |
| - !!python/tuple | |
| - 3.322785538319874 | |
| - 2.1624153548695157 | |
| - !!python/tuple | |
| - 3.489342008082259 | |
| - 2.469578333618206 | |
| - !!python/tuple | |
| - 3.278259980416589 | |
| - 1.6906831932492183 | |
| - !!python/tuple | |
| - 0.23045117210041233 | |
| - 0.26501940303452104 | |
| - !!python/tuple | |
| - 0.07818400556098729 | |
| - 0.027479091254378488 | |
| - !!python/tuple | |
| - 0.05445908159715249 | |
| - 0.013332869808672716 | |
| - !!python/tuple | |
| - 0.05042016806722689 | |
| - 0.0 | |
| - !!python/tuple | |
| - 0.05658536585365853 | |
| - 0.016585365853658534 | |
| - !!python/tuple | |
| - 0.05537665840439907 | |
| - 0.021182928316626336 | |
| - !!python/tuple | |
| - 0.04568795432892477 | |
| - 0.009015856562239487 | |
| - !!python/tuple | |
| - 0.047619047619047616 | |
| - 0.0 | |
| - !!python/tuple | |
| - 0.032679738562091505 | |
| - 0.0 | |
| - &id001 !!python/tuple | |
| - null | |
| - null | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| - *id001 | |
| length_ratio_std: 0.8476478699398606 | |
| max_observed_len_source: 153 | |
| max_observed_len_target: 87 | |
| num_discarded: 0 | |
| num_sents: 779247 | |
| num_sents_per_bucket: | |
| - 478128 | |
| - 233503 | |
| - 27823 | |
| - 9243 | |
| - 4982 | |
| - 3636 | |
| - 3802 | |
| - 4709 | |
| - 5275 | |
| - 4862 | |
| - 3208 | |
| - 32 | |
| - 8 | |
| - 12 | |
| - 2 | |
| - 4 | |
| - 8 | |
| - 6 | |
| - 2 | |
| - 2 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| - 0 | |
| num_tokens_source: 5305413 | |
| num_tokens_target: 7842731 | |
| num_unks_source: 0 | |
| num_unks_target: 0 | |
| size_vocab_source: 45864 | |
| size_vocab_target: 656 | |
| eop_id: -1 | |
| max_seq_len_source: 513 | |
| max_seq_len_target: 513 | |
| num_source_factors: 1 | |
| num_target_factors: 5 | |
| config_decoder: !TransformerConfig | |
| act_type: relu | |
| attention_heads: 8 | |
| block_prepended_cross_attention: false | |
| decoder_type: transformer | |
| depth_key_value: 512 | |
| dropout_act: 0.2 | |
| dropout_attention: 0.2 | |
| dropout_prepost: 0.2 | |
| feed_forward_num_hidden: 2048 | |
| max_seq_len_source: 513 | |
| max_seq_len_target: 513 | |
| model_size: 512 | |
| num_layers: 6 | |
| positional_embedding_type: fixed | |
| postprocess_sequence: dr | |
| preprocess_sequence: n | |
| use_glu: false | |
| use_lhuc: false | |
| config_embed_source: !EmbeddingConfig | |
| allow_sparse_grad: false | |
| dropout: 0.5 | |
| factor_configs: null | |
| num_embed: 512 | |
| num_factors: 1 | |
| vocab_size: 45864 | |
| config_embed_target: !EmbeddingConfig | |
| allow_sparse_grad: false | |
| dropout: 0.5 | |
| factor_configs: | |
| - !FactorConfig | |
| combine: sum | |
| num_embed: 512 | |
| share_embedding: false | |
| vocab_size: 16 | |
| - !FactorConfig | |
| combine: sum | |
| num_embed: 512 | |
| share_embedding: false | |
| vocab_size: 24 | |
| - !FactorConfig | |
| combine: sum | |
| num_embed: 512 | |
| share_embedding: false | |
| vocab_size: 416 | |
| - !FactorConfig | |
| combine: sum | |
| num_embed: 512 | |
| share_embedding: false | |
| vocab_size: 744 | |
| num_embed: 512 | |
| num_factors: 5 | |
| vocab_size: 656 | |
| config_encoder: !TransformerConfig | |
| act_type: relu | |
| attention_heads: 8 | |
| block_prepended_cross_attention: false | |
| decoder_type: transformer | |
| depth_key_value: 512 | |
| dropout_act: 0.2 | |
| dropout_attention: 0.2 | |
| dropout_prepost: 0.2 | |
| feed_forward_num_hidden: 2048 | |
| max_seq_len_source: 513 | |
| max_seq_len_target: 513 | |
| model_size: 512 | |
| num_layers: 6 | |
| positional_embedding_type: fixed | |
| postprocess_sequence: dr | |
| preprocess_sequence: n | |
| use_glu: false | |
| use_lhuc: false | |
| config_length_task: null | |
| dtype: float32 | |
| lhuc: false | |
| neural_vocab_selection: null | |
| neural_vocab_selection_block_loss: false | |
| vocab_source_size: 45864 | |
| vocab_target_size: 656 | |
| weight_tying_type: trg_softmax | |