[2025-07-05 21:26:12,625][__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: microsoft/phi-4 type: phi4_classification_lora num_labels: 6 output_dir: ./results/phi4-balanced/C2 logging_dir: ./logs/phi4-balanced/C2 best_model_dir: ./results/phi4-balanced/C2/best_model lora_r: 16 lora_dropout: 0.1 lora_alpha: 32 lora_target_modules: all-linear tokenizer: name: microsoft/phi-4 dataset: grade_index: 1 use_full_context: false training_params: weight_decay: 0.01 warmup_ratio: 0.1 learning_rate: 5.0e-05 train_batch_size: 8 eval_batch_size: 4 gradient_accumulation_steps: 2 gradient_checkpointing: true [2025-07-05 21:26:16,528][__main__][INFO] - GPU 0: NVIDIA H200 | TDP ≈ 700 W [2025-07-05 21:26:16,529][__main__][INFO] - Starting the Fine Tuning training process. [2025-07-05 21:26:20,036][transformers.tokenization_utils_base][INFO] - loading file vocab.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/vocab.json [2025-07-05 21:26:20,037][transformers.tokenization_utils_base][INFO] - loading file merges.txt from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/merges.txt [2025-07-05 21:26:20,037][transformers.tokenization_utils_base][INFO] - loading file tokenizer.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/tokenizer.json [2025-07-05 21:26:20,037][transformers.tokenization_utils_base][INFO] - loading file added_tokens.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/added_tokens.json [2025-07-05 21:26:20,037][transformers.tokenization_utils_base][INFO] - loading file special_tokens_map.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/special_tokens_map.json [2025-07-05 21:26:20,037][transformers.tokenization_utils_base][INFO] - loading file tokenizer_config.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/tokenizer_config.json [2025-07-05 21:26:20,037][transformers.tokenization_utils_base][INFO] - loading file chat_template.jinja from cache at None [2025-07-05 21:26:20,174][__main__][INFO] - Tokenizer function parameters- Padding:longest; Truncation: False; Use Full Context: False [2025-07-05 21:26:20,616][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 21:26:20,617][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "id2label": { "0": "LABEL_0", "1": "LABEL_1", "2": "LABEL_2", "3": "LABEL_3", "4": "LABEL_4", "5": "LABEL_5" }, "initializer_range": 0.02, "intermediate_size": 17920, "label2id": { "LABEL_0": 0, "LABEL_1": 1, "LABEL_2": 2, "LABEL_3": 3, "LABEL_4": 4, "LABEL_5": 5 }, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 21:26:20,742][transformers.modeling_utils][INFO] - loading weights file model.safetensors from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/model.safetensors.index.json [2025-07-05 21:26:20,742][transformers.modeling_utils][INFO] - Will use torch_dtype=torch.bfloat16 as defined in model's config object [2025-07-05 21:26:20,742][transformers.modeling_utils][INFO] - Instantiating Phi3ForSequenceClassification model under default dtype torch.bfloat16. [2025-07-05 21:26:38,885][transformers.modeling_utils][INFO] - Some weights of the model checkpoint at microsoft/phi-4 were not used when initializing Phi3ForSequenceClassification: ['lm_head.weight'] - This IS expected if you are initializing Phi3ForSequenceClassification 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 Phi3ForSequenceClassification from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model). [2025-07-05 21:26:38,886][transformers.modeling_utils][WARNING] - Some weights of Phi3ForSequenceClassification were not initialized from the model checkpoint at microsoft/phi-4 and are newly initialized: ['score.weight'] You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. [2025-07-05 21:26:40,771][__main__][INFO] - Initialized new PEFT model for ce loss [2025-07-05 21:26:40,773][__main__][INFO] - None [2025-07-05 21:26:40,774][transformers.training_args][INFO] - PyTorch: setting up devices [2025-07-05 21:26:40,856][__main__][INFO] - Total steps: 620. Number of warmup steps: 62 [2025-07-05 21:26:40,864][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. [2025-07-05 21:26:40,899][transformers.trainer][INFO] - Using auto half precision backend [2025-07-05 21:26:40,900][transformers.trainer][WARNING] - No label_names provided for model class `PeftModelForSequenceClassification`. Since `PeftModel` hides base models input arguments, if label_names is not given, label_names can't be set automatically within `Trainer`. Note that empty label_names list will be used instead. [2025-07-05 21:26:40,901][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 21:26:40,914][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 21:26:40,914][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 21:26:40,914][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 21:27:12,104][transformers.trainer][INFO] - The following columns in the Training set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 21:27:12,144][transformers.trainer][INFO] - ***** Running training ***** [2025-07-05 21:27:12,145][transformers.trainer][INFO] - Num examples = 500 [2025-07-05 21:27:12,145][transformers.trainer][INFO] - Num Epochs = 20 [2025-07-05 21:27:12,145][transformers.trainer][INFO] - Instantaneous batch size per device = 8 [2025-07-05 21:27:12,145][transformers.trainer][INFO] - Total train batch size (w. parallel, distributed & accumulation) = 16 [2025-07-05 21:27:12,145][transformers.trainer][INFO] - Gradient Accumulation steps = 2 [2025-07-05 21:27:12,145][transformers.trainer][INFO] - Total optimization steps = 640 [2025-07-05 21:27:12,147][transformers.trainer][INFO] - Number of trainable parameters = 55,736,320 [2025-07-05 21:27:12,220][transformers.models.phi3.modeling_phi3][WARNING] - `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`. [2025-07-05 21:34:38,541][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 21:34:38,544][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 21:34:38,544][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 21:34:38,544][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 21:35:09,548][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-32 [2025-07-05 21:35:09,883][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 21:35:09,883][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 21:42:36,801][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 21:42:36,805][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 21:42:36,805][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 21:42:36,805][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 21:43:07,804][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-64 [2025-07-05 21:43:08,157][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 21:43:08,157][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 21:43:09,192][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-32] due to args.save_total_limit [2025-07-05 21:50:19,725][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 21:50:19,728][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 21:50:19,728][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 21:50:19,728][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 21:50:50,713][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-96 [2025-07-05 21:50:51,028][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 21:50:51,029][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 21:58:02,670][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 21:58:02,674][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 21:58:02,674][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 21:58:02,674][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 21:58:33,656][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-128 [2025-07-05 21:58:34,139][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 21:58:34,140][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 21:58:35,264][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-64] due to args.save_total_limit [2025-07-05 21:58:35,306][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-96] due to args.save_total_limit [2025-07-05 22:05:46,315][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:05:46,319][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:05:46,319][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:05:46,319][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:06:17,331][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-160 [2025-07-05 22:06:17,656][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:06:17,657][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:06:18,681][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-128] due to args.save_total_limit [2025-07-05 22:13:29,603][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:13:29,606][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:13:29,606][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:13:29,606][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:14:00,611][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-192 [2025-07-05 22:14:00,972][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:14:00,973][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:21:12,840][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:21:12,844][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:21:12,844][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:21:12,844][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:21:43,865][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-224 [2025-07-05 22:21:44,191][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:21:44,192][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:21:45,176][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-160] due to args.save_total_limit [2025-07-05 22:21:45,213][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-192] due to args.save_total_limit [2025-07-05 22:28:55,981][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:28:55,985][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:28:55,985][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:28:55,985][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:29:26,956][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-256 [2025-07-05 22:29:27,278][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:29:27,278][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:29:28,287][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-224] due to args.save_total_limit [2025-07-05 22:36:39,084][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:36:39,087][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:36:39,087][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:36:39,087][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:37:10,080][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-288 [2025-07-05 22:37:10,429][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:37:10,430][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:44:22,288][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:44:22,292][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:44:22,292][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:44:22,292][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:44:53,305][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-320 [2025-07-05 22:44:53,792][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:44:53,793][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:44:54,798][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-288] due to args.save_total_limit [2025-07-05 22:52:05,331][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:52:05,334][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:52:05,334][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:52:05,334][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 22:52:36,297][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-352 [2025-07-05 22:52:36,623][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 22:52:36,623][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 22:52:37,743][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-320] due to args.save_total_limit [2025-07-05 22:59:48,239][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 22:59:48,242][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 22:59:48,242][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 22:59:48,242][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 23:00:19,238][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-384 [2025-07-05 23:00:19,605][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 23:00:19,606][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 23:00:20,591][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-352] due to args.save_total_limit [2025-07-05 23:07:31,162][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 23:07:31,165][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 23:07:31,165][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 23:07:31,165][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 23:08:02,186][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-416 [2025-07-05 23:08:02,509][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 23:08:02,509][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 23:08:03,583][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-384] due to args.save_total_limit [2025-07-05 23:08:03,624][transformers.trainer][INFO] - Training completed. Do not forget to share your model on huggingface.co/models =) [2025-07-05 23:08:03,624][transformers.trainer][INFO] - Loading best model from /workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-256 (score: 0.46039957188726366). [2025-07-05 23:08:03,965][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-05/21-26-12/results/phi4-balanced/C2/checkpoint-416] due to args.save_total_limit [2025-07-05 23:08:04,012][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 23:08:04,021][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 23:08:04,022][transformers.trainer][INFO] - Num examples = 132 [2025-07-05 23:08:04,022][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 23:08:35,010][__main__][INFO] - Training completed successfully. [2025-07-05 23:08:35,010][__main__][INFO] - Running on Test [2025-07-05 23:08:35,010][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt. If id, essay_text, reference, grades, supporting_text, id_prompt, essay_year, prompt are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-05 23:08:35,013][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-05 23:08:35,013][transformers.trainer][INFO] - Num examples = 138 [2025-07-05 23:08:35,013][transformers.trainer][INFO] - Batch size = 4 [2025-07-05 23:09:08,018][__main__][INFO] - Test metrics: {'eval_loss': 1.7057936191558838, 'eval_model_preparation_time': 0.0089, 'eval_accuracy': 0.2391304347826087, 'eval_RMSE': 72.87074123086873, 'eval_QWK': 0.20939657760432295, 'eval_HDIV': 0.14492753623188404, 'eval_Macro_F1': 0.1892660616537021, 'eval_Micro_F1': 0.2391304347826087, 'eval_Weighted_F1': 0.2273939139620625, 'eval_TP_0': 0, 'eval_TN_0': 137, 'eval_FP_0': 0, 'eval_FN_0': 1, 'eval_TP_1': 13, 'eval_TN_1': 62, 'eval_FP_1': 41, 'eval_FN_1': 22, 'eval_TP_2': 4, 'eval_TN_2': 103, 'eval_FP_2': 30, 'eval_FN_2': 1, 'eval_TP_3': 8, 'eval_TN_3': 71, 'eval_FP_3': 16, 'eval_FN_3': 43, 'eval_TP_4': 2, 'eval_TN_4': 108, 'eval_FP_4': 4, 'eval_FN_4': 24, 'eval_TP_5': 6, 'eval_TN_5': 104, 'eval_FP_5': 14, 'eval_FN_5': 14, 'eval_runtime': 32.9957, 'eval_samples_per_second': 4.182, 'eval_steps_per_second': 1.061, 'epoch': 13.0} [2025-07-05 23:09:08,019][transformers.trainer][INFO] - Saving model checkpoint to ./results/phi4-balanced/C2/best_model [2025-07-05 23:09:08,376][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-05 23:09:08,376][transformers.configuration_utils][INFO] - Model config Phi3Config { "architectures": [ "Phi3ForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 100257, "embd_pdrop": 0.0, "eos_token_id": 100265, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 17920, "max_position_embeddings": 16384, "model_type": "phi3", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 10, "original_max_position_embeddings": 16384, "pad_token_id": 100349, "partial_rotary_factor": 1.0, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 250000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.53.0", "use_cache": true, "vocab_size": 100352 } [2025-07-05 23:09:09,026][transformers.tokenization_utils_base][INFO] - chat template saved in ./results/phi4-balanced/C2/best_model/chat_template.jinja [2025-07-05 23:09:09,027][transformers.tokenization_utils_base][INFO] - tokenizer config file saved in ./results/phi4-balanced/C2/best_model/tokenizer_config.json [2025-07-05 23:09:09,027][transformers.tokenization_utils_base][INFO] - Special tokens file saved in ./results/phi4-balanced/C2/best_model/special_tokens_map.json [2025-07-05 23:09:09,098][__main__][INFO] - Model and tokenizer saved to ./results/phi4-balanced/C2/best_model [2025-07-05 23:09:09,113][__main__][INFO] - Fine Tuning Finished. [2025-07-05 23:09:09,622][__main__][INFO] - Total emissions: 0.0786 kg CO2eq