[2025-07-10 15:53:06,842][__main__][INFO] - cache_dir: /tmp/ dataset: name: kamel-usp/aes_enem_dataset split: JBCS2025 training_params: seed: 42 num_train_epochs: 20 logging_steps: 100 metric_for_best_model: QWK bf16: true bootstrap: enabled: true n_bootstrap: 10000 bootstrap_seed: 42 metrics: - QWK - Macro_F1 - Weighted_F1 post_training_results: model_path: /workspace/jbcs2025/outputs/2025-03-24/20-42-59 experiments: model: name: PORTULAN/albertina-1b5-portuguese-ptbr-encoder type: encoder_classification num_labels: 6 output_dir: ./results/ logging_dir: ./logs/ best_model_dir: ./results/best_model tokenizer: name: PORTULAN/albertina-1b5-portuguese-ptbr-encoder dataset: grade_index: 0 use_full_context: false training_params: weight_decay: 0.01 warmup_ratio: 0.1 learning_rate: 5.0e-05 train_batch_size: 4 eval_batch_size: 4 gradient_accumulation_steps: 4 gradient_checkpointing: false [2025-07-10 15:53:10,740][__main__][INFO] - GPU 0: NVIDIA RTX A6000 | TDP ≈ 300 W [2025-07-10 15:53:10,740][__main__][INFO] - Starting the Fine Tuning training process. [2025-07-10 15:53:15,881][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/config.json [2025-07-10 15:53:15,884][transformers.configuration_utils][INFO] - Model config DebertaV2Config { "architectures": [ "DebertaV2ForMaskedLM" ], "attention_head_size": 64, "attention_probs_dropout_prob": 0.1, "conv_act": "gelu", "conv_kernel_size": 3, "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 1536, "initializer_range": 0.02, "intermediate_size": 6144, "layer_norm_eps": 1e-07, "legacy": true, "max_position_embeddings": 512, "max_relative_positions": -1, "model_type": "deberta-v2", "norm_rel_ebd": "layer_norm", "num_attention_heads": 24, "num_hidden_layers": 48, "pad_token_id": 0, "pooler_dropout": 0, "pooler_hidden_act": "gelu", "pooler_hidden_size": 1536, "pos_att_type": [ "p2c", "c2p" ], "position_biased_input": false, "position_buckets": 256, "relative_attention": true, "share_att_key": true, "torch_dtype": "bfloat16", "transformers_version": "4.53.1", "type_vocab_size": 0, "vocab_size": 128100 } [2025-07-10 15:53:16,306][transformers.tokenization_utils_base][INFO] - loading file spm.model from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/spm.model [2025-07-10 15:53:16,307][transformers.tokenization_utils_base][INFO] - loading file tokenizer.json from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/tokenizer.json [2025-07-10 15:53:16,307][transformers.tokenization_utils_base][INFO] - loading file added_tokens.json from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/added_tokens.json [2025-07-10 15:53:16,307][transformers.tokenization_utils_base][INFO] - loading file special_tokens_map.json from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/special_tokens_map.json [2025-07-10 15:53:16,307][transformers.tokenization_utils_base][INFO] - loading file tokenizer_config.json from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/tokenizer_config.json [2025-07-10 15:53:16,307][transformers.tokenization_utils_base][INFO] - loading file chat_template.jinja from cache at None [2025-07-10 15:53:16,570][__main__][INFO] - Tokenizer function parameters- Padding:longest; Truncation: True; Use Full Context: False [2025-07-10 15:53:16,972][__main__][INFO] - Token statistics for 'train' split: [2025-07-10 15:53:16,972][__main__][INFO] - Total examples: 500 [2025-07-10 15:53:16,972][__main__][INFO] - Min tokens: 512 [2025-07-10 15:53:16,972][__main__][INFO] - Max tokens: 512 [2025-07-10 15:53:16,972][__main__][INFO] - Avg tokens: 512.00 [2025-07-10 15:53:16,972][__main__][INFO] - Std tokens: 0.00 [2025-07-10 15:53:17,067][__main__][INFO] - Token statistics for 'validation' split: [2025-07-10 15:53:17,067][__main__][INFO] - Total examples: 132 [2025-07-10 15:53:17,067][__main__][INFO] - Min tokens: 512 [2025-07-10 15:53:17,067][__main__][INFO] - Max tokens: 512 [2025-07-10 15:53:17,067][__main__][INFO] - Avg tokens: 512.00 [2025-07-10 15:53:17,067][__main__][INFO] - Std tokens: 0.00 [2025-07-10 15:53:17,166][__main__][INFO] - Token statistics for 'test' split: [2025-07-10 15:53:17,166][__main__][INFO] - Total examples: 138 [2025-07-10 15:53:17,167][__main__][INFO] - Min tokens: 512 [2025-07-10 15:53:17,167][__main__][INFO] - Max tokens: 512 [2025-07-10 15:53:17,167][__main__][INFO] - Avg tokens: 512.00 [2025-07-10 15:53:17,167][__main__][INFO] - Std tokens: 0.00 [2025-07-10 15:53:17,167][__main__][INFO] - If token statistics are the same (max, avg, min) keep in mind that this is due to batched tokenization and padding. [2025-07-10 15:53:17,167][__main__][INFO] - Model max length: 512. If it is the same as stats, then there is a high chance that sequences are being truncated. [2025-07-10 15:53:17,431][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/config.json [2025-07-10 15:53:17,432][transformers.configuration_utils][INFO] - Model config DebertaV2Config { "architectures": [ "DebertaV2ForMaskedLM" ], "attention_head_size": 64, "attention_probs_dropout_prob": 0.1, "conv_act": "gelu", "conv_kernel_size": 3, "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 1536, "id2label": { "0": 0, "1": 40, "2": 80, "3": 120, "4": 160, "5": 200 }, "initializer_range": 0.02, "intermediate_size": 6144, "label2id": { "0": 0, "40": 1, "80": 2, "120": 3, "160": 4, "200": 5 }, "layer_norm_eps": 1e-07, "legacy": true, "max_position_embeddings": 512, "max_relative_positions": -1, "model_type": "deberta-v2", "norm_rel_ebd": "layer_norm", "num_attention_heads": 24, "num_hidden_layers": 48, "pad_token_id": 0, "pooler_dropout": 0, "pooler_hidden_act": "gelu", "pooler_hidden_size": 1536, "pos_att_type": [ "p2c", "c2p" ], "position_biased_input": false, "position_buckets": 256, "relative_attention": true, "share_att_key": true, "torch_dtype": "bfloat16", "transformers_version": "4.53.1", "type_vocab_size": 0, "vocab_size": 128100 } [2025-07-10 15:53:17,820][transformers.modeling_utils][INFO] - loading weights file pytorch_model.bin from cache at /tmp/models--PORTULAN--albertina-1b5-portuguese-ptbr-encoder/snapshots/b22008e5096af9c398b75762d4e28e5008762916/pytorch_model.bin [2025-07-10 15:53:17,821][transformers.modeling_utils][INFO] - Will use torch_dtype=torch.bfloat16 as defined in model's config object [2025-07-10 15:53:17,821][transformers.modeling_utils][INFO] - Instantiating DebertaV2ForSequenceClassification model under default dtype torch.bfloat16. [2025-07-10 15:53:18,201][transformers.safetensors_conversion][INFO] - Attempting to create safetensors variant [2025-07-10 15:53:18,665][transformers.safetensors_conversion][INFO] - Attempting to convert .bin model on the fly to safetensors. [2025-07-10 16:02:29,595][transformers.modeling_utils][INFO] - Some weights of the model checkpoint at PORTULAN/albertina-1b5-portuguese-ptbr-encoder were not used when initializing DebertaV2ForSequenceClassification: ['cls.predictions.bias', 'cls.predictions.decoder.bias', 'cls.predictions.decoder.weight', 'cls.predictions.transform.LayerNorm.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.transform.dense.weight'] - This IS expected if you are initializing DebertaV2ForSequenceClassification from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model). - This IS NOT expected if you are initializing DebertaV2ForSequenceClassification from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model). [2025-07-10 16:02:29,595][transformers.modeling_utils][WARNING] - Some weights of DebertaV2ForSequenceClassification were not initialized from the model checkpoint at PORTULAN/albertina-1b5-portuguese-ptbr-encoder and are newly initialized: ['classifier.bias', 'classifier.weight', 'pooler.dense.bias', 'pooler.dense.weight'] You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. [2025-07-10 16:02:29,613][transformers.training_args][INFO] - PyTorch: setting up devices [2025-07-10 16:02:29,639][__main__][INFO] - Total steps: 620. Number of warmup steps: 62 [2025-07-10 16:02:29,649][transformers.trainer][INFO] - You have loaded a model on multiple GPUs. `is_model_parallel` attribute will be force-set to `True` to avoid any unexpected behavior such as device placement mismatching. [2025-07-10 16:02:29,676][transformers.trainer][INFO] - Using auto half precision backend [2025-07-10 16:02:29,678][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:02:29,689][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:02:29,689][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:02:29,690][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:02:38,845][transformers.trainer][INFO] - The following columns in the Training set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:02:38,876][transformers.trainer][INFO] - ***** Running training ***** [2025-07-10 16:02:38,876][transformers.trainer][INFO] - Num examples = 500 [2025-07-10 16:02:38,876][transformers.trainer][INFO] - Num Epochs = 20 [2025-07-10 16:02:38,876][transformers.trainer][INFO] - Instantaneous batch size per device = 4 [2025-07-10 16:02:38,876][transformers.trainer][INFO] - Total train batch size (w. parallel, distributed & accumulation) = 16 [2025-07-10 16:02:38,876][transformers.trainer][INFO] - Gradient Accumulation steps = 4 [2025-07-10 16:02:38,876][transformers.trainer][INFO] - Total optimization steps = 640 [2025-07-10 16:02:38,878][transformers.trainer][INFO] - Number of trainable parameters = 1,566,919,686 [2025-07-10 16:09:32,605][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:09:32,609][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:09:32,609][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:09:32,609][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:09:41,083][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-32 [2025-07-10 16:09:41,085][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-32/config.json [2025-07-10 16:09:46,899][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-32/model.safetensors [2025-07-10 16:16:50,105][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:16:50,109][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:16:50,110][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:16:50,110][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:16:58,598][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-64 [2025-07-10 16:16:58,601][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-64/config.json [2025-07-10 16:17:05,052][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-64/model.safetensors [2025-07-10 16:17:13,233][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-32] due to args.save_total_limit [2025-07-10 16:24:09,684][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:24:09,688][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:24:09,688][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:24:09,688][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:24:18,146][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-96 [2025-07-10 16:24:18,148][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-96/config.json [2025-07-10 16:24:23,608][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-96/model.safetensors [2025-07-10 16:24:30,971][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-64] due to args.save_total_limit [2025-07-10 16:31:27,837][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:31:27,841][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:31:27,842][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:31:27,842][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:31:36,301][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-128 [2025-07-10 16:31:36,304][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-128/config.json [2025-07-10 16:31:43,337][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-128/model.safetensors [2025-07-10 16:31:49,743][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-96] due to args.save_total_limit [2025-07-10 16:38:52,376][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:38:52,380][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:38:52,380][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:38:52,380][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:39:00,857][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-160 [2025-07-10 16:39:00,862][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-160/config.json [2025-07-10 16:39:10,224][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-160/model.safetensors [2025-07-10 16:46:16,187][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:46:16,213][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:46:16,214][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:46:16,214][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:46:24,732][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-192 [2025-07-10 16:46:24,736][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-192/config.json [2025-07-10 16:46:31,956][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-192/model.safetensors [2025-07-10 16:46:38,914][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-160] due to args.save_total_limit [2025-07-10 16:53:35,545][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 16:53:35,550][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 16:53:35,550][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 16:53:35,550][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 16:53:44,013][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-224 [2025-07-10 16:53:44,015][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-224/config.json [2025-07-10 16:53:48,622][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-224/model.safetensors [2025-07-10 16:53:55,055][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-192] due to args.save_total_limit [2025-07-10 17:00:51,991][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 17:00:51,995][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 17:00:51,995][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 17:00:51,995][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 17:01:00,444][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-256 [2025-07-10 17:01:00,447][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-256/config.json [2025-07-10 17:01:05,603][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-256/model.safetensors [2025-07-10 17:01:12,787][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-224] due to args.save_total_limit [2025-07-10 17:08:09,449][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 17:08:09,456][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 17:08:09,456][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 17:08:09,456][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 17:08:17,939][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-288 [2025-07-10 17:08:17,941][transformers.configuration_utils][INFO] - Configuration saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-288/config.json [2025-07-10 17:08:24,232][transformers.modeling_utils][INFO] - Model weights saved in /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-288/model.safetensors [2025-07-10 17:08:33,143][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-256] due to args.save_total_limit [2025-07-10 17:08:33,643][transformers.trainer][INFO] - Training completed. Do not forget to share your model on huggingface.co/models =) [2025-07-10 17:08:33,644][transformers.trainer][INFO] - Loading best model from /workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-128 (score: 0.6863342898134864). [2025-07-10 17:08:36,030][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-10/15-53-06/results/checkpoint-288] due to args.save_total_limit [2025-07-10 17:08:36,692][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 17:08:36,696][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 17:08:36,696][transformers.trainer][INFO] - Num examples = 132 [2025-07-10 17:08:36,696][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 17:08:45,160][__main__][INFO] - Training completed successfully. [2025-07-10 17:08:45,160][__main__][INFO] - Running on Test [2025-07-10 17:08:45,161][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `DebertaV2ForSequenceClassification.forward` and have been ignored: prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades. If prompt, id, essay_year, id_prompt, supporting_text, reference, essay_text, grades are not expected by `DebertaV2ForSequenceClassification.forward`, you can safely ignore this message. [2025-07-10 17:08:45,164][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-10 17:08:45,164][transformers.trainer][INFO] - Num examples = 138 [2025-07-10 17:08:45,164][transformers.trainer][INFO] - Batch size = 4 [2025-07-10 17:08:53,928][__main__][INFO] - Test metrics: {'eval_loss': 0.9303792119026184, 'eval_model_preparation_time': 0.0074, 'eval_accuracy': 0.7028985507246377, 'eval_RMSE': 25.931906372573962, 'eval_QWK': 0.6826328310864394, 'eval_HDIV': 0.007246376811594235, 'eval_Macro_F1': 0.49688027796692447, 'eval_Micro_F1': 0.7028985507246377, 'eval_Weighted_F1': 0.7116672747290037, 'eval_TP_0': 0, 'eval_TN_0': 137, 'eval_FP_0': 0, 'eval_FN_0': 1, 'eval_TP_1': 0, 'eval_TN_1': 138, 'eval_FP_1': 0, 'eval_FN_1': 0, 'eval_TP_2': 6, 'eval_TN_2': 122, 'eval_FP_2': 6, 'eval_FN_2': 4, 'eval_TP_3': 45, 'eval_TN_3': 65, 'eval_FP_3': 7, 'eval_FN_3': 21, 'eval_TP_4': 40, 'eval_TN_4': 71, 'eval_FP_4': 16, 'eval_FN_4': 11, 'eval_TP_5': 6, 'eval_TN_5': 116, 'eval_FP_5': 12, 'eval_FN_5': 4, 'eval_runtime': 8.7537, 'eval_samples_per_second': 15.765, 'eval_steps_per_second': 3.998, 'epoch': 9.0} [2025-07-10 17:08:53,929][transformers.trainer][INFO] - Saving model checkpoint to ./results/best_model [2025-07-10 17:08:53,931][transformers.configuration_utils][INFO] - Configuration saved in ./results/best_model/config.json [2025-07-10 17:08:58,786][transformers.modeling_utils][INFO] - Model weights saved in ./results/best_model/model.safetensors [2025-07-10 17:08:58,789][transformers.tokenization_utils_base][INFO] - tokenizer config file saved in ./results/best_model/tokenizer_config.json [2025-07-10 17:08:58,789][transformers.tokenization_utils_base][INFO] - Special tokens file saved in ./results/best_model/special_tokens_map.json [2025-07-10 17:08:58,816][__main__][INFO] - Model and tokenizer saved to ./results/best_model [2025-07-10 17:08:58,940][__main__][INFO] - Fine Tuning Finished. [2025-07-10 17:08:59,455][__main__][INFO] - Total emissions: 0.1036 kg CO2eq