[2025-07-04 10:03:14,713][__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 tokenizer: name: microsoft/phi-4 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: 8 eval_batch_size: 4 gradient_accumulation_steps: 2 gradient_checkpointing: true [2025-07-04 10:03:18,601][__main__][INFO] - GPU 0: NVIDIA H200 | TDP ≈ 700 W [2025-07-04 10:03:18,602][__main__][INFO] - Starting the Fine Tuning training process. [2025-07-04 10:03:22,406][transformers.tokenization_utils_base][INFO] - loading file vocab.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/vocab.json [2025-07-04 10:03:22,406][transformers.tokenization_utils_base][INFO] - loading file merges.txt from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/merges.txt [2025-07-04 10:03:22,407][transformers.tokenization_utils_base][INFO] - loading file tokenizer.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/tokenizer.json [2025-07-04 10:03:22,407][transformers.tokenization_utils_base][INFO] - loading file added_tokens.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/added_tokens.json [2025-07-04 10:03:22,407][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 [2025-07-04 10:03:22,407][transformers.tokenization_utils_base][INFO] - loading file tokenizer_config.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/tokenizer_config.json [2025-07-04 10:03:22,407][transformers.tokenization_utils_base][INFO] - loading file chat_template.jinja from cache at None [2025-07-04 10:03:22,543][__main__][INFO] - Tokenizer function parameters- Padding:longest; Truncation: False; Use Full Context: False [2025-07-04 10:03:23,670][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:03:23,671][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 } [2025-07-04 10:03:23,796][transformers.modeling_utils][INFO] - loading weights file model.safetensors from cache at /tmp/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/model.safetensors.index.json [2025-07-04 10:03:23,797][transformers.modeling_utils][INFO] - Will use torch_dtype=torch.bfloat16 as defined in model's config object [2025-07-04 10:03:23,797][transformers.modeling_utils][INFO] - Instantiating Phi3ForSequenceClassification model under default dtype torch.bfloat16. [2025-07-04 10:03:42,475][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). [2025-07-04 10:03:42,476][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. [2025-07-04 10:03:43,669][__main__][INFO] - Initialized new PEFT model for ce loss [2025-07-04 10:03:43,671][__main__][INFO] - None [2025-07-04 10:03:43,673][transformers.training_args][INFO] - PyTorch: setting up devices [2025-07-04 10:03:43,749][__main__][INFO] - Total steps: 620. Number of warmup steps: 62 [2025-07-04 10:03:43,757][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. [2025-07-04 10:03:43,791][transformers.trainer][INFO] - Using auto half precision backend [2025-07-04 10:03:43,791][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. [2025-07-04 10:03:43,792][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:03:43,805][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:03:43,805][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:03:43,805][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:04:05,303][transformers.trainer][INFO] - The following columns in the Training set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:04:05,342][transformers.trainer][INFO] - ***** Running training ***** [2025-07-04 10:04:05,342][transformers.trainer][INFO] - Num examples = 500 [2025-07-04 10:04:05,342][transformers.trainer][INFO] - Num Epochs = 20 [2025-07-04 10:04:05,342][transformers.trainer][INFO] - Instantaneous batch size per device = 8 [2025-07-04 10:04:05,342][transformers.trainer][INFO] - Total train batch size (w. parallel, distributed & accumulation) = 16 [2025-07-04 10:04:05,342][transformers.trainer][INFO] - Gradient Accumulation steps = 2 [2025-07-04 10:04:05,342][transformers.trainer][INFO] - Total optimization steps = 640 [2025-07-04 10:04:05,344][transformers.trainer][INFO] - Number of trainable parameters = 27,883,520 [2025-07-04 10:04:05,400][transformers.models.phi3.modeling_phi3][WARNING] - `use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`. [2025-07-04 10:09:03,841][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:09:03,844][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:09:03,844][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:09:03,844][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:09:25,044][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-32 [2025-07-04 10:09:25,516][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:09:25,517][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 } [2025-07-04 10:14:25,014][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:14:25,017][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:14:25,017][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:14:25,017][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:14:46,203][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-64 [2025-07-04 10:14:46,535][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:14:46,535][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 } [2025-07-04 10:14:47,215][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-32] due to args.save_total_limit [2025-07-04 10:19:46,105][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:19:46,109][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:19:46,109][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:19:46,109][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:20:07,290][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-96 [2025-07-04 10:20:07,611][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:20:07,612][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 } [2025-07-04 10:20:08,156][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-64] due to args.save_total_limit [2025-07-04 10:25:07,108][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:25:07,111][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:25:07,112][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:25:07,112][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:25:28,287][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-128 [2025-07-04 10:25:28,791][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:25:28,792][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 } [2025-07-04 10:25:29,354][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-96] due to args.save_total_limit [2025-07-04 10:30:28,196][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:30:28,199][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:30:28,199][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:30:28,199][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:30:49,355][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-160 [2025-07-04 10:30:49,875][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:30:49,876][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 } [2025-07-04 10:35:50,440][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:35:50,444][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:35:50,444][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:35:50,444][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:36:11,658][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-192 [2025-07-04 10:36:12,139][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:36:12,140][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 } [2025-07-04 10:36:12,719][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-128] due to args.save_total_limit [2025-07-04 10:36:12,739][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-160] due to args.save_total_limit [2025-07-04 10:41:12,367][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:41:12,371][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:41:12,371][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:41:12,371][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:41:33,601][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-224 [2025-07-04 10:41:33,924][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:41:33,924][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 } [2025-07-04 10:41:34,480][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-192] due to args.save_total_limit [2025-07-04 10:46:34,006][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:46:34,010][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:46:34,010][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:46:34,010][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:46:55,235][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-256 [2025-07-04 10:46:55,567][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:46:55,567][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 } [2025-07-04 10:51:55,806][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:51:55,809][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:51:55,810][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:51:55,810][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:52:17,026][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-288 [2025-07-04 10:52:17,345][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:52:17,346][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 } [2025-07-04 10:52:17,922][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-256] due to args.save_total_limit [2025-07-04 10:57:17,323][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 10:57:17,327][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 10:57:17,327][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 10:57:17,327][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 10:57:38,533][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-320 [2025-07-04 10:57:38,854][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 10:57:38,855][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 } [2025-07-04 10:57:39,432][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-224] due to args.save_total_limit [2025-07-04 10:57:39,452][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-288] due to args.save_total_limit [2025-07-04 11:02:38,729][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:02:38,733][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:02:38,733][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 11:02:38,733][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:02:59,930][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-352 [2025-07-04 11:03:00,295][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 11:03:00,296][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 } [2025-07-04 11:08:00,095][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:08:00,099][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:08:00,099][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 11:08:00,099][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:08:21,309][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-384 [2025-07-04 11:08:21,656][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 11:08:21,657][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 } [2025-07-04 11:08:22,308][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-352] due to args.save_total_limit [2025-07-04 11:13:21,039][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:13:21,043][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:13:21,043][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 11:13:21,043][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:13:42,174][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-416 [2025-07-04 11:13:42,504][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 11:13:42,505][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 } [2025-07-04 11:13:43,050][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-384] due to args.save_total_limit [2025-07-04 11:18:40,950][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:18:40,953][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:18:40,953][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 11:18:40,953][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:19:02,065][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-448 [2025-07-04 11:19:02,380][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 11:19:02,380][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 } [2025-07-04 11:19:02,918][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-416] due to args.save_total_limit [2025-07-04 11:24:01,109][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:24:01,112][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:24:01,113][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 11:24:01,113][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:24:22,214][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-480 [2025-07-04 11:24:22,543][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 11:24:22,543][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 } [2025-07-04 11:24:23,207][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-448] due to args.save_total_limit [2025-07-04 11:24:23,220][transformers.trainer][INFO] - Training completed. Do not forget to share your model on huggingface.co/models =) [2025-07-04 11:24:23,220][transformers.trainer][INFO] - Loading best model from /workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-320 (score: 0.7337395027169438). [2025-07-04 11:24:23,654][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-07-04/10-03-14/results/checkpoint-480] due to args.save_total_limit [2025-07-04 11:24:23,674][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:24:23,680][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:24:23,680][transformers.trainer][INFO] - Num examples = 132 [2025-07-04 11:24:23,680][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:24:44,750][__main__][INFO] - Training completed successfully. [2025-07-04 11:24:44,750][__main__][INFO] - Running on Test [2025-07-04 11:24:44,750][transformers.trainer][INFO] - The following columns in the Evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades. If prompt, essay_text, essay_year, id, id_prompt, supporting_text, reference, grades are not expected by `PeftModelForSequenceClassification.forward`, you can safely ignore this message. [2025-07-04 11:24:44,753][transformers.trainer][INFO] - ***** Running Evaluation ***** [2025-07-04 11:24:44,753][transformers.trainer][INFO] - Num examples = 138 [2025-07-04 11:24:44,753][transformers.trainer][INFO] - Batch size = 4 [2025-07-04 11:25:07,305][__main__][INFO] - Test metrics: {'eval_loss': 1.0189533233642578, 'eval_model_preparation_time': 0.009, 'eval_accuracy': 0.6159420289855072, 'eval_RMSE': 27.240208984279956, 'eval_QWK': 0.6675700090334237, 'eval_HDIV': 0.007246376811594235, 'eval_Macro_F1': 0.4546094428447369, 'eval_Micro_F1': 0.6159420289855072, 'eval_Weighted_F1': 0.6116426909265785, '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': 9, 'eval_TN_2': 113, 'eval_FP_2': 15, 'eval_FN_2': 1, 'eval_TP_3': 30, 'eval_TN_3': 66, 'eval_FP_3': 6, 'eval_FN_3': 36, 'eval_TP_4': 41, 'eval_TN_4': 62, 'eval_FP_4': 25, 'eval_FN_4': 10, 'eval_TP_5': 5, 'eval_TN_5': 121, 'eval_FP_5': 7, 'eval_FN_5': 5, 'eval_runtime': 22.5429, 'eval_samples_per_second': 6.122, 'eval_steps_per_second': 1.553, 'epoch': 15.0} [2025-07-04 11:25:07,306][transformers.trainer][INFO] - Saving model checkpoint to ./results/best_model [2025-07-04 11:25:07,626][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /workspace/.hf_home/hub/models--microsoft--phi-4/snapshots/187ef0342fff0eb3333be9f00389385e95ef0b61/config.json [2025-07-04 11:25:07,626][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 } [2025-07-04 11:25:07,974][transformers.tokenization_utils_base][INFO] - chat template saved in ./results/best_model/chat_template.jinja [2025-07-04 11:25:07,975][transformers.tokenization_utils_base][INFO] - tokenizer config file saved in ./results/best_model/tokenizer_config.json [2025-07-04 11:25:07,975][transformers.tokenization_utils_base][INFO] - Special tokens file saved in ./results/best_model/special_tokens_map.json [2025-07-04 11:25:08,043][__main__][INFO] - Model and tokenizer saved to ./results/best_model [2025-07-04 11:25:08,065][__main__][INFO] - Fine Tuning Finished. [2025-07-04 11:25:08,574][__main__][INFO] - Total emissions: 0.0631 kg CO2eq