[2025-06-30 11:58:45,575][__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/ logging_dir: ./logs/ best_model_dir: ./results/best_model lora_r: 8 lora_dropout: 0.05 lora_alpha: 16 lora_target_modules: all-linear checkpoint_path: '' tokenizer: name: microsoft/phi-4 dataset: grade_index: 4 use_full_context: true training_params: weight_decay: 0.01 warmup_ratio: 0.1 learning_rate: 5.0e-05 train_batch_size: 1 eval_batch_size: 16 gradient_accumulation_steps: 16 gradient_checkpointing: false [2025-06-30 11:58:50,444][__main__][INFO] - GPU 0: NVIDIA H200 | TDP ≈ 700 W [2025-06-30 11:58:50,444][__main__][INFO] - Starting the Fine Tuning training process. [2025-06-30 11:58:56,163][transformers.tokenization_utils_base][INFO] - loading file vocab.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/vocab.json [2025-06-30 11:58:56,164][transformers.tokenization_utils_base][INFO] - loading file merges.txt from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/merges.txt [2025-06-30 11:58:56,164][transformers.tokenization_utils_base][INFO] - loading file tokenizer.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/tokenizer.json [2025-06-30 11:58:56,164][transformers.tokenization_utils_base][INFO] - loading file added_tokens.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/added_tokens.json [2025-06-30 11:58:56,164][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 [2025-06-30 11:58:56,164][transformers.tokenization_utils_base][INFO] - loading file tokenizer_config.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/tokenizer_config.json [2025-06-30 11:58:56,164][transformers.tokenization_utils_base][INFO] - loading file chat_template.jinja from cache at None [2025-06-30 11:58:56,308][__main__][INFO] - Tokenizer function parameters- Padding:longest; Truncation: False; Use Full Context: True [2025-06-30 11:58:58,114][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 11:58:58,115][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 } [2025-06-30 11:58:58,252][transformers.modeling_utils][INFO] - loading weights file model.safetensors from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/model.safetensors.index.json [2025-06-30 11:58:58,252][transformers.modeling_utils][INFO] - Will use torch_dtype=torch.bfloat16 as defined in model's config object [2025-06-30 11:58:58,252][transformers.modeling_utils][INFO] - Instantiating Phi3ForSequenceClassification model under default dtype torch.bfloat16. [2025-06-30 11:59:04,705][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). [2025-06-30 11:59:04,705][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. [2025-06-30 11:59:06,000][__main__][INFO] - Initialized new PEFT model for ce loss [2025-06-30 11:59:06,002][__main__][INFO] - None [2025-06-30 11:59:06,003][transformers.training_args][INFO] - PyTorch: setting up devices [2025-06-30 11:59:06,064][__main__][INFO] - Total steps: 620. Number of warmup steps: 62 [2025-06-30 11:59:06,069][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. [2025-06-30 11:59:06,094][transformers.trainer][INFO] - Using auto half precision backend [2025-06-30 11:59:06,095][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. [2025-06-30 11:59:06,096][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 11:59:06,109][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 11:59:06,109][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 11:59:06,109][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 11:59:47,004][transformers.trainer][INFO] - The following columns in the Training set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 11:59:47,044][transformers.trainer][INFO] - ***** Running training ***** [2025-06-30 11:59:47,044][transformers.trainer][INFO] - Num examples = 500 [2025-06-30 11:59:47,044][transformers.trainer][INFO] - Num Epochs = 20 [2025-06-30 11:59:47,044][transformers.trainer][INFO] - Instantaneous batch size per device = 1 [2025-06-30 11:59:47,044][transformers.trainer][INFO] - Total train batch size (w. parallel, distributed & accumulation) = 16 [2025-06-30 11:59:47,044][transformers.trainer][INFO] - Gradient Accumulation steps = 16 [2025-06-30 11:59:47,044][transformers.trainer][INFO] - Total optimization steps = 640 [2025-06-30 11:59:47,046][transformers.trainer][INFO] - Number of trainable parameters = 27,883,520 [2025-06-30 12:05:59,762][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:05:59,766][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:05:59,766][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:05:59,766][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:06:40,394][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-32 [2025-06-30 12:06:40,875][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:06:40,875][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 } [2025-06-30 12:12:54,216][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:12:54,220][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:12:54,220][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:12:54,220][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:13:34,768][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-64 [2025-06-30 12:13:35,532][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:13:35,533][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 } [2025-06-30 12:13:36,029][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-32] due to args.save_total_limit [2025-06-30 12:19:49,229][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:19:49,233][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:19:49,233][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:19:49,233][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:20:29,840][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-96 [2025-06-30 12:20:30,289][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:20:30,290][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 } [2025-06-30 12:26:47,323][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:26:47,326][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:26:47,327][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:26:47,327][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:27:27,864][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-128 [2025-06-30 12:27:28,387][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:27:28,388][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 } [2025-06-30 12:27:28,939][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-96] due to args.save_total_limit [2025-06-30 12:33:42,531][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:33:42,534][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:33:42,534][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:33:42,534][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:34:22,995][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-160 [2025-06-30 12:34:23,487][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:34:23,488][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 } [2025-06-30 12:34:24,108][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-64] due to args.save_total_limit [2025-06-30 12:34:24,126][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-128] due to args.save_total_limit [2025-06-30 12:40:37,239][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:40:37,242][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:40:37,242][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:40:37,243][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:41:17,780][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-192 [2025-06-30 12:41:18,287][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:41:18,287][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 } [2025-06-30 12:41:19,017][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-160] due to args.save_total_limit [2025-06-30 12:47:31,796][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:47:31,799][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:47:31,799][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:47:31,799][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:48:12,364][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-224 [2025-06-30 12:48:12,987][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:48:12,988][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 } [2025-06-30 12:48:13,636][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-192] due to args.save_total_limit [2025-06-30 12:54:26,770][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 12:54:26,773][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 12:54:26,773][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 12:54:26,773][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 12:55:07,349][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-256 [2025-06-30 12:55:07,943][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 12:55:07,944][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 } [2025-06-30 12:55:08,631][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-224] due to args.save_total_limit [2025-06-30 13:01:21,467][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:01:21,470][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:01:21,470][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:01:21,470][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:02:02,020][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-288 [2025-06-30 13:02:02,498][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:02:02,498][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 } [2025-06-30 13:02:03,139][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-256] due to args.save_total_limit [2025-06-30 13:08:16,526][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:08:16,530][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:08:16,530][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:08:16,530][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:08:57,122][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-320 [2025-06-30 13:08:57,628][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:08:57,629][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 } [2025-06-30 13:15:11,839][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:15:11,842][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:15:11,842][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:15:11,842][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:15:52,416][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-352 [2025-06-30 13:15:52,945][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:15:52,946][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 } [2025-06-30 13:15:53,621][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-320] due to args.save_total_limit [2025-06-30 13:22:07,209][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:22:07,212][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:22:07,212][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:22:07,213][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:22:47,773][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-384 [2025-06-30 13:22:48,322][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:22:48,323][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 } [2025-06-30 13:22:48,836][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-352] due to args.save_total_limit [2025-06-30 13:29:01,917][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:29:01,921][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:29:01,921][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:29:01,921][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:29:42,573][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-416 [2025-06-30 13:29:43,069][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:29:43,069][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 } [2025-06-30 13:29:43,730][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-288] due to args.save_total_limit [2025-06-30 13:29:43,756][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-384] due to args.save_total_limit [2025-06-30 13:35:57,179][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:35:57,183][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:35:57,183][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:35:57,183][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:36:37,696][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-448 [2025-06-30 13:36:38,228][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:36:38,229][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 } [2025-06-30 13:42:52,757][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:42:52,760][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:42:52,760][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:42:52,760][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:43:33,299][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-480 [2025-06-30 13:43:33,793][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:43:33,794][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 } [2025-06-30 13:43:34,489][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-448] due to args.save_total_limit [2025-06-30 13:49:47,514][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:49:47,517][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:49:47,517][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:49:47,517][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:50:28,031][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-512 [2025-06-30 13:50:28,527][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:50:28,528][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 } [2025-06-30 13:50:29,127][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-480] due to args.save_total_limit [2025-06-30 13:56:42,214][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 13:56:42,217][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 13:56:42,217][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 13:56:42,217][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 13:57:22,845][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-544 [2025-06-30 13:57:23,316][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 13:57:23,317][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 } [2025-06-30 13:57:23,932][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-512] due to args.save_total_limit [2025-06-30 14:03:37,084][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 14:03:37,087][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 14:03:37,087][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 14:03:37,087][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 14:04:17,556][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-576 [2025-06-30 14:04:18,058][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 14:04:18,058][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 } [2025-06-30 14:04:18,728][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-544] due to args.save_total_limit [2025-06-30 14:04:18,748][transformers.trainer][INFO] - Training completed. Do not forget to share your model on huggingface.co/models =) [2025-06-30 14:04:18,748][transformers.trainer][INFO] - Loading best model from /workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-416 (score: 0.6509961790393013). [2025-06-30 14:04:18,895][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-06-30/11-58-45/results/checkpoint-576] due to args.save_total_limit [2025-06-30 14:04:18,920][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 14:04:18,926][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 14:04:18,926][transformers.trainer][INFO] - Num examples = 132 [2025-06-30 14:04:18,926][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 14:04:59,479][__main__][INFO] - Training completed successfully. [2025-06-30 14:04:59,479][__main__][INFO] - Running on Test [2025-06-30 14:04:59,479][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year. If grades, id, supporting_text, prompt, reference, id_prompt, essay_text, essay_year are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-06-30 14:04:59,483][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-06-30 14:04:59,483][transformers.trainer][INFO] - Num examples = 138 [2025-06-30 14:04:59,483][transformers.trainer][INFO] - Batch size = 16 [2025-06-30 14:05:41,361][__main__][INFO] - Test metrics: {'eval_loss': 1.6296720504760742, 'eval_model_preparation_time': 0.0097, 'eval_accuracy': 0.36231884057971014, 'eval_RMSE': 56.568542494923804, 'eval_QWK': 0.5073723420766724, 'eval_HDIV': 0.05797101449275366, 'eval_Macro_F1': 0.3552263104501911, 'eval_Micro_F1': 0.36231884057971014, 'eval_Weighted_F1': 0.3672380004905022, 'eval_TP_0': 11, 'eval_TN_0': 111, 'eval_FP_0': 5, 'eval_FN_0': 11, 'eval_TP_1': 16, 'eval_TN_1': 78, 'eval_FP_1': 28, 'eval_FN_1': 16, 'eval_TP_2': 4, 'eval_TN_2': 106, 'eval_FP_2': 8, 'eval_FN_2': 20, 'eval_TP_3': 8, 'eval_TN_3': 79, 'eval_FP_3': 34, 'eval_FN_3': 17, 'eval_TP_4': 10, 'eval_TN_4': 96, 'eval_FP_4': 10, 'eval_FN_4': 22, 'eval_TP_5': 1, 'eval_TN_5': 132, 'eval_FP_5': 3, 'eval_FN_5': 2, 'eval_runtime': 41.8688, 'eval_samples_per_second': 3.296, 'eval_steps_per_second': 0.215, 'epoch': 18.0} [2025-06-30 14:05:41,362][transformers.trainer][INFO] - Saving model checkpoint to ./results/best_model [2025-06-30 14:05:41,899][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-06-30 14:05:41,900][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 } [2025-06-30 14:05:42,311][transformers.tokenization_utils_base][INFO] - chat template saved in ./results/best_model/chat_template.jinja [2025-06-30 14:05:42,312][transformers.tokenization_utils_base][INFO] - tokenizer config file saved in ./results/best_model/tokenizer_config.json [2025-06-30 14:05:42,312][transformers.tokenization_utils_base][INFO] - Special tokens file saved in ./results/best_model/special_tokens_map.json [2025-06-30 14:05:42,386][__main__][INFO] - Model and tokenizer saved to ./results/best_model [2025-06-30 14:05:42,411][__main__][INFO] - Fine Tuning Finished. [2025-06-30 14:05:42,918][__main__][INFO] - Total emissions: 0.0039 kg CO2eq