model_args.json

{"adafactor_beta1": null, "adafactor_clip_threshold": 1.0, "adafactor_decay_rate": -0.8, "adafactor_eps": [1e-30, 0.001], "adafactor_relative_step": true, "adafactor_scale_parameter": true, "adafactor_warmup_init": true, "adam_betas": [0.9, 0.999], "adam_epsilon": 1e-08, "best_model_dir": "outputs/best_model", "cache_dir": "cache_dir/", "config": {}, "cosine_schedule_num_cycles": 0.5, "custom_layer_parameters": [], "custom_parameter_groups": [], "dataloader_num_workers": 0, "do_lower_case": false, "dynamic_quantize": false, "early_stopping_consider_epochs": false, "early_stopping_delta": 0, "early_stopping_metric": "eval_loss", "early_stopping_metric_minimize": true, "early_stopping_patience": 3, "encoding": null, "eval_batch_size": 100, "evaluate_during_training": false, "evaluate_during_training_silent": true, "evaluate_during_training_steps": 2000, "evaluate_during_training_verbose": false, "evaluate_each_epoch": true, "fp16": true, "gradient_accumulation_steps": 1, "learning_rate": 4e-05, "local_rank": -1, "logging_steps": 50, "loss_type": null, "loss_args": {}, "manual_seed": 64, "max_grad_norm": 1.0, "max_seq_length": 512, "model_name": "classla/bcms-bertic", "model_type": "electra", "multiprocessing_chunksize": -1, "n_gpu": 1, "no_cache": true, "no_save": false, "not_saved_args": [], "num_train_epochs": 20, "optimizer": "AdamW", "output_dir": "outputs/", "overwrite_output_dir": true, "polynomial_decay_schedule_lr_end": 1e-07, "polynomial_decay_schedule_power": 1.0, "process_count": 14, "quantized_model": false, "reprocess_input_data": true, "save_best_model": true, "save_eval_checkpoints": false, "save_model_every_epoch": false, "save_optimizer_and_scheduler": true, "save_steps": 2000, "scheduler": "linear_schedule_with_warmup", "silent": true, "skip_special_tokens": true, "tensorboard_dir": null, "thread_count": null, "tokenizer_name": null, "tokenizer_type": null, "train_batch_size": 8, "train_custom_parameters_only": false, "trust_remote_code": false, "use_cached_eval_features": false, "use_early_stopping": false, "use_hf_datasets": false, "use_multiprocessing": true, "use_multiprocessing_for_evaluation": true, "wandb_kwargs": {}, "wandb_project": null, "warmup_ratio": 0.06, "warmup_steps": 716, "weight_decay": 0.0, "model_class": "NERModel", "classification_report": false, "labels_list": ["[0, '', 0, '']", "[1, 'evar', 0, '']", "[1, 'a', 0, '']", "[1, '', 0, '']", "[1, 'oj', 0, '']", "[2, '', 0, '']", "[6, 'ao', 0, '']", "[3, 'ivati', 0, '']", "[0, 'ti', 0, '']", "[0, 'be', 0, '']", "[2, 'an', 0, '']", "[4, 'ao', 0, '']", "[1, 'i', 0, '']", "[2, 'biti', 0, '']", "[2, 'iti', 0, '']", "[2, 'aj', 0, '']", "[1, 'e', 0, '']", "[3, 'an', 0, '']", "[2, 'ar', 0, '']", "[2, 'ji', 0, '']", "[1, 'eti', 0, '']", "[2, 'ti', 0, '']", "[3, 'lac', 0, '']", "[1, 'iti', 0, '']", "[1, 'o', 0, '']", "[2, 'av', 0, '']", "[3, 'iti', 0, '']", "[4, 'ti', 0, '']", "[4, 'iti', 0, '']", "[1, 'ji', 0, '']", "[3, '', 0, '']", "[2, 'ak', 0, '']", "[6, 'ak', 0, '']", "[5, '', 0, '']", "[3, 'o', 0, '']", "[3, 'aj', 0, '']", "[0, 'ti', 1, 'htj']", "[2, 'ti', 1, 'htj']", "[3, 'ak', 0, '']", "[2, 'i', 0, '']", "[3, 'ovati', 0, '']", "[1, 'sti', 0, '']", "[3, 'ka', 0, '']", "[2, 'a', 0, '']", "[4, 'li', 0, '']", "[2, '\u0107i', 0, '']", "[3, 'av', 0, '']", "[4, 'k', 0, '']", "[0, 'j', 0, '']", "[3, 'e', 0, '']", "[4, 'li', 3, '']", "[3, '', 0, 'o']", "[1, 'aj', 0, '']", "[2, 'eti', 0, '']", "[3, 'lik', 3, '']", "[2, 'zati', 0, '']", "[4, 'biti', 0, '']", "[1, '', 0, 'on']", "[1, 'ti', 0, '']", "[2, 'ti', 1, 'b']", "[5, 'iti', 0, '']", "[1, 'j', 0, '']", "[2, 'ka', 0, '']", "[1, 'jeti', 0, '']", "[5, 'ivati', 0, '']", "[4, 'ak', 0, '']", "[2, 'vati', 0, '']", "[3, 'st', 0, '']", "[3, 'jeti', 0, '']", "[3, 'ek', 0, '']", "[0, 'a', 0, '']", "[2, 'at', 0, '']", "[2, 'ac', 0, '']", "[1, 'ica', 0, '']", "[1, 'i\u010dan', 0, '']", "[1, 'arta', 0, '']", "[2, 'k', 0, '']", "[4, 'o', 0, '']", "[1, 'astarski', 0, '']", "[1, 'ojan', 0, '']", "[6, 'o', 0, '']", "[4, 'ac', 0, '']", "[3, 'o', 3, '']", "[2, 'o', 0, '']", "[4, 'io', 0, '']", "[3, 'vo', 0, '']", "[2, 'cati', 0, '']", "[1, 'ati', 0, '']", "[2, 'tak', 0, '']", "[2, 'on', 0, '']", "[2, 'ao', 0, '']", "[3, '\u0107i', 0, '']", "[3, 'ac', 0, '']", "[3, 'ar', 0, '']", "[4, '', 0, 'o']", "[3, 't', 0, '']", "[2, 'sok', 0, '']", "[2, '.o.o.', 0, '']", "[3, 'lo', 0, '']", "[3, 'ati', 0, '']", "[3, 'biti', 0, '']", "[2, 'g', 0, '']", "[5, 'an', 3, '']", "[1, 'istarski', 0, '']", "[3, 'tan', 0, '']", "[2, '', 2, 'vrij']", "[0, '', 3, '\u0161t']", "[4, '\u0161ta', 0, '']", "[2, 'jeti', 0, '']", "[2, 'sti', 0, '']", "[5, '\u0107i', 0, '']", "[3, 'a', 0, '']", "[1, 'an', 0, '']", "[2, 'lik', 0, '']", "[1, 'rokuplje', 0, '']", "[1, 'epublika', 0, '']", "[3, 'sti', 0, '']", "[2, 'eko', 0, '']", "[4, 'diti', 0, '']", "[2, 'ga', 0, '']", "[4, 'lac', 0, '']", "[0, 'i', 0, '']", "[4, 'av', 0, '']", "[2, 'zu', 0, '']", "[3, 'zak', 0, '']", "[1, 'ije', 0, '']", "[4, 'titi', 0, '']", "[3, 'tak', 3, '']", "[4, 'vati', 0, '']", "[3, 'jeta', 0, '']", "[3, 'lo', 3, '']", "[3, 'ijeti', 0, '']", "[3, 'ba', 0, '']", "[1, 'u\u017eben', 0, '']", "[1, 'ugi', 0, '']", "[4, 'at', 0, '']", "[0, '', 1, 'sj']", "[3, 'titi', 0, '']", "[2, 'sati', 0, '']", "[4, 'djeti', 0, '']", "[5, 'stiti', 0, '']", "[1, 'a', 1, 'sj']", "[2, 'a', 1, 'sj']", "[5, 'o', 0, '']", "[6, 'ogo', 0, 'm']", "[0, '', 2, 'prij']", "[3, 'djeti', 0, '']", "[4, 'an', 0, '']", "[2, 'tan', 0, '']", "[2, 'gati', 0, '']", "[1, 'i', 3, 'bi']", "[4, 'mnogo', 0, '']", "[4, 'g', 0, '']", "[4, '', 0, '']", "[5, 'ijeti', 0, '']", "[2, 'ek', 0, '']", "[2, 'nuti', 0, '']", "[5, 'lac', 0, '']", "[1, 'uti', 0, '']", "[3, 'jena', 0, '']", "[3, 'li', 0, '']", "[0, '', 4, 'procj']", "[6, 'jeljivati', 0, '']", "[1, '', 4, 'procij']", "[1, 'e', 2, 'odj']", "[0, '', 2, 'odj']", "[2, 'jev', 0, '']", "[5, 'ijeliti', 0, '']", "[1, 'i\u010dno', 0, '']", "[6, 'r', 1, 'dob']", "[3, 'iti', 1, 'cij']", "[4, '\u0161to', 0, '']", "[2, 'zak', 0, '']", "[1, 'r\u0161ilac', 0, '']", "[1, 'u\u017enost', 0, '']", "[1, 'av', 0, '']", "[4, 'am', 0, '']", "[3, 'diti', 0, '']", "[4, 'tak', 0, '']", "[2, 'ha', 0, '']", "[5, 'slati', 0, '']", "[3, 'lik', 0, '']", "[1, 'ok', 0, '']", "[3, '', 2, 'vrij']", "[4, 'i', 0, 'o']", "[2, 'go', 0, '']", "[5, 'jetan', 0, '']", "[2, '', 1, 'di']", "[2, 'te', 1, 'di']", "[1, 'ak', 0, '']", "[4, 'ziti', 0, '']", "[2, 'tac', 0, '']", "[4, 'an', 3, '']", "[1, 'a', 1, 'cij']", "[2, 'tko', 0, '']", "[2, 'zak', 3, '']", "[3, 'sok', 0, '']", "[3, 'ao', 0, '']", "[3, 'g', 0, '']", "[3, 'e', 2, 'odj']", "[4, 'ena', 0, '']", "[2, 'sko', 0, '']", "[5, 'jednji', 0, '']", "[0, '', 1, 'mj']", "[7, 'ijepiti', 0, '']", "[2, 'tati', 0, '']", "[4, '\u0107i', 0, '']", "[1, 'a\u010dka', 0, '']", "[1, 'a', 1, 'mj']", "[2, 'eko', 3, '']", "[3, 'i', 0, 'o']", "[8, 'o', 0, '']", "[2, 'go', 3, '']", "[3, 'avati', 0, '']", "[2, '', 0, 'o']", "[3, 'siti', 0, '']", "[5, 'je\u010dan', 0, '']", "[1, 'agistar', 0, '']", "[6, 'iti', 0, '']", "[4, 'citi', 0, '']", "[4, 't', 0, '']", "[3, 'iti', 1, 'rij']", "[3, 'vojeru\u010dno', 0, '']", "[4, 'gati', 0, '']", "[4, 'stiti', 0, '']", "[3, 'ti', 0, '']", "[1, 'odina', 0, '']", "[1, '\u0161i', 0, '']", "[3, 'jer', 0, '']", "[1, 'u\u0107an', 0, '']", "[1, 'e', 3, 'v']", "[1, 'resuda', 0, '']", "[1, 'p\u0161tinski', 0, '']", "[1, 'ud', 0, '']", "[2, 'an', 5, 'prav']", "[3, 'ijebiti', 0, '']", "[1, 'okat', 0, '']", "[1, 'en', 0, '']", "[1, 'oljetan', 0, '']", "[3, 'ta', 0, '']", "[3, 'tac', 0, '']", "[2, 'am', 0, '']", "[1, 'izija', 0, '']", "[4, 'lja', 0, '']", "[5, 'jinski', 1, 'dij']", "[1, 'omo\u0107nik', 0, '']", "[2, 'j', 0, '']", "[7, 'r', 1, 'dob']", "[3, '', 1, 'di']", "[8, 'eniti', 0, '']", "[1, 'ka', 0, '']", "[3, 'i', 0, '']", "[5, 'ijeme', 0, '']", "[7, 'eniti', 0, '']", "[4, 'eo', 0, '']", "[5, 'ki', 0, '']", "[1, 'n', 1, '']", "[1, 'ar\u0161al', 0, '']", "[5, 'k', 1, 'uz']", "[1, 'ovodan', 0, '']", "[4, 'sti', 0, '']", "[5, 'ovati', 0, '']", "[3, 'ajati', 0, '']", "[3, 'jev', 0, '']", "[3, 'an', 2, 'm']", "[3, 'zac', 0, '']", "[5, 'eniti', 0, '']", "[3, 'jeriti', 0, '']", "[6, 'eljan', 0, '']", "[0, 'k', 0, '']", "[2, '', 5, 'isposa']", "[4, 'a', 0, '']", "[5, 'eo', 0, '']", "[3, 'je', 0, '']", "[4, 'tati', 0, '']", "[1, 'ana', 0, '']", "[2, 'htjeti', 0, '']", "[3, 'tjeti', 0, '']", "[1, 'esor', 0, '']", "[5, 'estan', 0, '']", "[3, 'li', 3, '']", "[0, 'ar', 0, '']", "[2, 'oktor', 0, '']", "[3, 'ek', 1, '\u010dov']", "[3, 'ti', 1, 'b']", "[3, 'anj', 0, '']", "[4, 'siti', 0, '']", "[2, 'lik', 3, '']", "[1, 'e', 1, 'vij']", "[1, 'uriranje', 0, '']", "[0, '', 3, 'r']", "[1, 'i', 2, 'n']", "[5, 'sti', 0, '']", "[3, 'dak', 0, '']", "[6, 'stiti', 0, '']", "[0, 'ugi', 0, '']", "[4, 'ar', 0, 'do']", "[3, 'ijediti', 0, '']", "[4, 'ivati', 0, '']", "[2, 'ja', 0, '']", "[1, 'amostalno', 0, '']", "[1, 'reduze\u0107e', 0, '']", "[0, '', 1, 'dj']", "[2, 'zo', 0, '']", "[5, 'io', 0, '']", "[2, 'ti', 3, 'izvj']", "[2, 'n', 0, '']", "[5, 'an', 0, '']", "[0, 'ti', 3, 'izvj']", "[1, 'in', 0, '']", "[3, 'at', 0, '']", "[3, 'slan', 0, '']", "[6, 'ar', 0, '']", "[5, 'ent', 0, '']", "[3, 'ti', 1, 'sl']", "[3, 'tak', 0, '']", "[6, 'ed', 0, '']", "[1, 'osno', 0, '']", "[4, 'cati', 0, '']", "[3, 'mo', 0, '']", "[1, 'efon', 0, '']", "[4, 'jedba', 0, '']", "[1, 'ugo', 0, '']", "[1, 'omiran', 0, '']", "[2, 'io', 0, '']", "[4, 'zak', 0, '']", "[0, '', 4, 'bezbj']", "[1, 'ina', 0, '']", "[6, 'jeljan', 0, '']", "[1, '', 2, 'c']", "[3, 'vati', 0, '']", "[1, 'oktor', 0, '']", "[1, '', 2, 'n']", "[0, 'an', 0, '']", "[0, 'ica', 0, '']", "[3, '\u0161ta', 0, '']", "[4, 'ti', 2, '']", "[2, 'ti', 4, 'inten']", "[0, '.', 0, '']", "[2, 'sok', 3, '']", "[3, 'ovati', 1, 'dj']", "[0, 'botica', 0, '']", "[1, 'ektri\u010dan', 0, '']", "[1, 'i', 2, '']", "[0, 'ti', 1, 'b']", "[1, 'ena', 0, '']", "[2, '', 5, 'vodod']", "[5, 'zniti', 0, '']", "[0, 'n', 0, '']", "[2, 'g', 3, '']", "[5, 'jeriti', 0, '']", "[4, 'len', 0, '']", "[1, 'omo\u0107je', 0, '']", "[2, 'ati', 0, '']", "[1, 'ospodin', 0, '']", "[5, 'jeniti', 0, '']", "[2, '', 11, 'prvotu\u017een']", "[4, 'dak', 0, '']", "[1, '', 5, 'zaht']", "[3, 'len', 0, '']", "[5, 'ijetiti', 0, '']", "[0, '', 4, 'zahti']", "[3, 'ijek', 0, '']", "[0, 'ti', 3, 'podi']", "[2, 'ok', 0, '']"], "lazy_loading": false, "lazy_loading_start_line": 0, "onnx": false, "special_tokens_list": []}