Pushing fine-tuned model to Hugging Face Hub

README.md

@@ -0,0 +1,43 @@
+
+---
+language:
+- pt
+- en
+tags:
+- aes
+datasets:
+- kamel-usp/aes_enem_dataset
+base_model: meta-llama/Llama-3.1-8B
+metrics:
+- accuracy
+- qwk
+library_name: peft
+model-index:
+  - name: llama31_8b-balanced-C1
+    results:
+      - task:
+          type: text-classification
+          name: Automated Essay Score
+        dataset:
+          name: Automated Essay Score ENEM Dataset
+          type: kamel-usp/aes_enem_dataset
+          config: JBCS2025
+          split: test
+        metrics:
+          - name: Macro F1
+            type: F1
+            value: 0.4357443349423437
+          - name: QWK
+            type: qwk
+            value: 0.6043890865954924
+---
+# Model ID: llama31_8b-balanced-C1
+## Results
+|               |   test_data |
+|:--------------|------------:|
+| eval_accuracy |  0.65942    |
+| eval_RMSE     | 25.9319     |
+| eval_QWK      |  0.604389   |
+| eval_Macro_F1 |  0.435744   |
+| eval_HDIV     |  0.00724638 |
+        

adapter_config.json

@@ -0,0 +1,40 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "meta-llama/Llama-3.1-8B",
+  "bias": "none",
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 16,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": [
+    "classifier",
+    "score"
+  ],
+  "peft_type": "LORA",
+  "r": 8,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "o_proj",
+    "gate_proj",
+    "v_proj",
+    "down_proj",
+    "up_proj",
+    "q_proj",
+    "k_proj"
+  ],
+  "task_type": "SEQ_CLS",
+  "use_dora": false,
+  "use_rslora": false
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6397979039118a3634b35deb1598ac9506c0cc7ceca748624eca7a92dbeec49
+size 83994544

run_experiment.log

@@ -0,0 +1,1147 @@
+[2025-03-16 13:15:47,890][__main__][INFO] - cache_dir: /media/data/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
+post_training_results:
+  model_path: /workspace/jbcs2025/outputs/2025-03-16/08-36-36
+experiments:
+  model:
+    name: meta-llama/Llama-3.1-8B
+    type: llama31_classification_lora
+    num_labels: 6
+    output_dir: ./results/llama31_8b-balanced/C1
+    logging_dir: ./logs/llama31_8b-balanced/C1
+    best_model_dir: ./results/llama31_8b-balanced/C1/best_model
+    lora_r: 8
+    lora_dropout: 0.05
+    lora_alpha: 16
+    lora_target_modules: all-linear
+  dataset:
+    grade_index: 0
+  training_id: llama31_8b-balanced-C1
+  training_params:
+    weight_decay: 0.01
+    warmup_ratio: 0.1
+    learning_rate: 5.0e-05
+    train_batch_size: 1
+    eval_batch_size: 2
+    gradient_accumulation_steps: 16
+    gradient_checkpointing: false
+
+[2025-03-16 13:15:47,891][__main__][INFO] - Starting the Fine Tuning training process.
+[2025-03-16 13:15:51,975][transformers.tokenization_utils_base][INFO] - loading file tokenizer.json from cache at /media/data/tmp/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/tokenizer.json
+[2025-03-16 13:15:51,975][transformers.tokenization_utils_base][INFO] - loading file tokenizer.model from cache at None
+[2025-03-16 13:15:51,975][transformers.tokenization_utils_base][INFO] - loading file added_tokens.json from cache at None
+[2025-03-16 13:15:51,975][transformers.tokenization_utils_base][INFO] - loading file special_tokens_map.json from cache at /media/data/tmp/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/special_tokens_map.json
+[2025-03-16 13:15:51,975][transformers.tokenization_utils_base][INFO] - loading file tokenizer_config.json from cache at /media/data/tmp/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/tokenizer_config.json
+[2025-03-16 13:15:51,975][transformers.tokenization_utils_base][INFO] - loading file chat_template.jinja from cache at None
+[2025-03-16 13:15:52,167][transformers.tokenization_utils_base][INFO] - Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.
+[2025-03-16 13:15:52,172][__main__][INFO] - Tokenizer function parameters- Padding:longest; Truncation: False
+[2025-03-16 13:15:52,783][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /media/data/tmp/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 13:15:52,786][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "_name_or_path": "meta-llama/Llama-3.1-8B",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "id2label": {
+    "0": 0,
+    "1": 40,
+    "2": 80,
+    "3": 120,
+    "4": 160,
+    "5": 200
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "label2id": {
+    "0": 0,
+    "40": 1,
+    "80": 2,
+    "120": 3,
+    "160": 4,
+    "200": 5
+  },
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 13:15:52,800][transformers.modeling_utils][INFO] - loading weights file model.safetensors from cache at /media/data/tmp/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/model.safetensors.index.json
+[2025-03-16 13:15:52,801][transformers.modeling_utils][INFO] - Will use torch_dtype=torch.bfloat16 as defined in model's config object
+[2025-03-16 13:15:52,801][transformers.modeling_utils][INFO] - Instantiating LlamaForSequenceClassification model under default dtype torch.bfloat16.
+[2025-03-16 13:15:56,600][transformers.modeling_utils][INFO] - Some weights of the model checkpoint at meta-llama/Llama-3.1-8B were not used when initializing LlamaForSequenceClassification: {'lm_head.weight'}
+- This IS expected if you are initializing LlamaForSequenceClassification 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 LlamaForSequenceClassification from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).
+[2025-03-16 13:15:56,601][transformers.modeling_utils][WARNING] - Some weights of LlamaForSequenceClassification were not initialized from the model checkpoint at meta-llama/Llama-3.1-8B 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.
+[2025-03-16 13:15:56,832][__main__][INFO] - None
+[2025-03-16 13:15:56,833][transformers.training_args][INFO] - PyTorch: setting up devices
+[2025-03-16 13:15:56,871][__main__][INFO] - Total steps: 620. Number of warmup steps: 62
+[2025-03-16 13:15:56,880][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.
+[2025-03-16 13:15:56,893][transformers.trainer][INFO] - Using auto half precision backend
+[2025-03-16 13:15:56,893][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.
+[2025-03-16 13:15:56,912][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 13:15:56,925][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 13:15:56,925][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 13:15:56,925][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 13:16:58,143][transformers][INFO] - {'accuracy': 0.29545454545454547, 'RMSE': 46.056618647183825, 'QWK': 0.03250125649187485, 'HDIV': 0.015151515151515138, 'Macro F1': 0.07692307692307693}
+[2025-03-16 13:16:58,145][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 13:16:58,289][transformers.trainer][INFO] - The following columns in the training set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] - ***** Running training *****
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] -   Num examples = 500
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] -   Num Epochs = 20
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] -   Instantaneous batch size per device = 1
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] -   Total train batch size (w. parallel, distributed & accumulation) = 16
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] -   Gradient Accumulation steps = 16
+[2025-03-16 13:16:58,325][transformers.trainer][INFO] -   Total optimization steps = 620
+[2025-03-16 13:16:58,328][transformers.trainer][INFO] -   Number of trainable parameters = 20,996,096
+[2025-03-16 13:30:53,496][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 13:30:53,497][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 13:30:53,497][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 13:30:53,498][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 13:31:57,120][transformers][INFO] - {'accuracy': 0.3939393939393939, 'RMSE': 41.046905910780715, 'QWK': 0.07965489566613149, 'HDIV': 0.007575757575757569, 'Macro F1': 0.1734968771164121}
+[2025-03-16 13:31:57,120][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 13:31:57,122][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-32
+[2025-03-16 13:31:57,604][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 13:31:57,606][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 13:45:58,066][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 13:45:58,067][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 13:45:58,067][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 13:45:58,068][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 13:47:01,924][transformers][INFO] - {'accuracy': 0.4696969696969697, 'RMSE': 33.574882386580704, 'QWK': 0.25437317784256563, 'HDIV': 0.007575757575757569, 'Macro F1': 0.19694989106753813}
+[2025-03-16 13:47:01,924][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 13:47:01,926][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-64
+[2025-03-16 13:47:02,387][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 13:47:02,390][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 13:47:02,559][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-32] due to args.save_total_limit
+[2025-03-16 14:01:02,989][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 14:01:02,991][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 14:01:02,991][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 14:01:02,991][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 14:02:06,865][transformers][INFO] - {'accuracy': 0.4772727272727273, 'RMSE': 32.84490643597388, 'QWK': 0.3022095509622237, 'HDIV': 0.007575757575757569, 'Macro F1': 0.2089608257095942}
+[2025-03-16 14:02:06,866][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 14:02:06,867][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-96
+[2025-03-16 14:02:07,358][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 14:02:07,360][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 14:02:07,530][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-64] due to args.save_total_limit
+[2025-03-16 14:16:07,909][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 14:16:07,910][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 14:16:07,910][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 14:16:07,910][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 14:17:11,741][transformers][INFO] - {'accuracy': 0.45454545454545453, 'RMSE': 38.92494720807615, 'QWK': 0.2276727204643325, 'HDIV': 0.007575757575757569, 'Macro F1': 0.30417862838915466}
+[2025-03-16 14:17:11,741][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 14:17:11,742][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-128
+[2025-03-16 14:17:12,265][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 14:17:12,267][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 14:31:12,254][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 14:31:12,256][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 14:31:12,256][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 14:31:12,256][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 14:32:16,023][transformers][INFO] - {'accuracy': 0.5, 'RMSE': 33.574882386580704, 'QWK': 0.16648560564910375, 'HDIV': 0.007575757575757569, 'Macro F1': 0.183216929010737}
+[2025-03-16 14:32:16,023][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 14:32:16,025][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-160
+[2025-03-16 14:32:16,566][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 14:32:16,568][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 14:32:16,736][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-128] due to args.save_total_limit
+[2025-03-16 14:46:17,000][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 14:46:17,001][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 14:46:17,001][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 14:46:17,001][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 14:47:20,843][transformers][INFO] - {'accuracy': 0.5984848484848485, 'RMSE': 28.91995221924885, 'QWK': 0.5161495962600935, 'HDIV': 0.015151515151515138, 'Macro F1': 0.2924253285543608}
+[2025-03-16 14:47:20,843][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 14:47:20,844][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-192
+[2025-03-16 14:47:21,310][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 14:47:21,313][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 14:47:21,478][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-96] due to args.save_total_limit
+[2025-03-16 14:47:21,490][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-160] due to args.save_total_limit
+[2025-03-16 15:01:21,920][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 15:01:21,921][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 15:01:21,921][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 15:01:21,921][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 15:02:25,838][transformers][INFO] - {'accuracy': 0.5757575757575758, 'RMSE': 32.84490643597388, 'QWK': 0.276957163958641, 'HDIV': 0.007575757575757569, 'Macro F1': 0.2738722188201674}
+[2025-03-16 15:02:25,838][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 15:02:25,839][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-224
+[2025-03-16 15:02:26,368][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 15:02:26,370][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 15:16:27,446][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 15:16:27,447][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 15:16:27,447][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 15:16:27,447][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 15:17:31,369][transformers][INFO] - {'accuracy': 0.5909090909090909, 'RMSE': 29.336088024923512, 'QWK': 0.5172058520502782, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3985035687797945}
+[2025-03-16 15:17:31,369][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 15:17:31,371][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-256
+[2025-03-16 15:17:31,878][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 15:17:31,881][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 15:17:32,051][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-192] due to args.save_total_limit
+[2025-03-16 15:17:32,063][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-224] due to args.save_total_limit
+[2025-03-16 15:31:34,746][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 15:31:34,747][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 15:31:34,747][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 15:31:34,747][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 15:32:39,461][transformers][INFO] - {'accuracy': 0.5833333333333334, 'RMSE': 28.91995221924885, 'QWK': 0.4639830508474576, 'HDIV': 0.007575757575757569, 'Macro F1': 0.28377402226355086}
+[2025-03-16 15:32:39,462][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 15:32:39,463][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-288
+[2025-03-16 15:32:39,927][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 15:32:39,929][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 15:46:43,172][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 15:46:43,174][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 15:46:43,174][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 15:46:43,174][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 15:47:47,303][transformers][INFO] - {'accuracy': 0.5833333333333334, 'RMSE': 27.633971188310298, 'QWK': 0.510593220338983, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3399487967229903}
+[2025-03-16 15:47:47,303][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 15:47:47,304][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-320
+[2025-03-16 15:47:47,769][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 15:47:47,771][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 15:47:47,938][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-288] due to args.save_total_limit
+[2025-03-16 16:01:51,129][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 16:01:51,130][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 16:01:51,130][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 16:01:51,130][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 16:02:55,323][transformers][INFO] - {'accuracy': 0.5984848484848485, 'RMSE': 27.19179912021158, 'QWK': 0.5295629820051413, 'HDIV': 0.007575757575757569, 'Macro F1': 0.34896996773392897}
+[2025-03-16 16:02:55,323][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 16:02:55,324][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-352
+[2025-03-16 16:02:55,847][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 16:02:55,849][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 16:02:56,019][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-256] due to args.save_total_limit
+[2025-03-16 16:02:56,031][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-320] due to args.save_total_limit
+[2025-03-16 16:17:00,650][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 16:17:00,651][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 16:17:00,651][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 16:17:00,651][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 16:18:05,272][transformers][INFO] - {'accuracy': 0.553030303030303, 'RMSE': 30.944720996896347, 'QWK': 0.4280386134269418, 'HDIV': 0.007575757575757569, 'Macro F1': 0.27696997064800183}
+[2025-03-16 16:18:05,273][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 16:18:05,274][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-384
+[2025-03-16 16:18:05,734][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 16:18:05,736][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 16:32:09,622][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 16:32:09,623][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 16:32:09,623][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 16:32:09,623][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 16:33:13,955][transformers][INFO] - {'accuracy': 0.5833333333333334, 'RMSE': 28.91995221924885, 'QWK': 0.49003359462486007, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3452991452991453}
+[2025-03-16 16:33:13,955][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 16:33:13,956][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-416
+[2025-03-16 16:33:14,443][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 16:33:14,445][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 16:33:14,611][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-384] due to args.save_total_limit
+[2025-03-16 16:47:19,993][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 16:47:19,994][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 16:47:19,994][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 16:47:19,994][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 16:48:24,194][transformers][INFO] - {'accuracy': 0.5227272727272727, 'RMSE': 32.84490643597388, 'QWK': 0.4285992217898833, 'HDIV': 0.007575757575757569, 'Macro F1': 0.36892246283550634}
+[2025-03-16 16:48:24,194][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 16:48:24,196][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-448
+[2025-03-16 16:48:24,730][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 16:48:24,732][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 16:48:24,895][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-416] due to args.save_total_limit
+[2025-03-16 17:02:29,963][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 17:02:29,964][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 17:02:29,964][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 17:02:29,964][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 17:03:34,106][transformers][INFO] - {'accuracy': 0.6136363636363636, 'RMSE': 26.74231693686086, 'QWK': 0.5616839261593878, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3411934470758}
+[2025-03-16 17:03:34,106][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 17:03:34,107][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-480
+[2025-03-16 17:03:34,596][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 17:03:34,598][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 17:03:34,762][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-352] due to args.save_total_limit
+[2025-03-16 17:03:34,774][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-448] due to args.save_total_limit
+[2025-03-16 17:17:38,703][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 17:17:38,704][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 17:17:38,704][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 17:17:38,704][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 17:18:43,041][transformers][INFO] - {'accuracy': 0.5606060606060606, 'RMSE': 30.15113445777636, 'QWK': 0.49084550504011537, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3855266713094572}
+[2025-03-16 17:18:43,042][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 17:18:43,043][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-512
+[2025-03-16 17:18:43,576][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 17:18:43,578][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 17:32:49,532][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 17:32:49,533][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 17:32:49,533][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 17:32:49,533][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 17:33:53,775][transformers][INFO] - {'accuracy': 0.5909090909090909, 'RMSE': 29.336088024923512, 'QWK': 0.5077731092436977, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3968880268150341}
+[2025-03-16 17:33:53,776][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 17:33:53,777][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-544
+[2025-03-16 17:33:54,264][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 17:33:54,266][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 17:33:54,431][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-512] due to args.save_total_limit
+[2025-03-16 17:47:58,410][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 17:47:58,412][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 17:47:58,412][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 17:47:58,412][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 17:49:02,705][transformers][INFO] - {'accuracy': 0.6287878787878788, 'RMSE': 26.285149626910837, 'QWK': 0.579476861167002, 'HDIV': 0.007575757575757569, 'Macro F1': 0.394676583276989}
+[2025-03-16 17:49:02,706][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 17:49:02,707][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-576
+[2025-03-16 17:49:03,279][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 17:49:03,281][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 17:49:03,447][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-480] due to args.save_total_limit
+[2025-03-16 17:49:03,459][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-544] due to args.save_total_limit
+[2025-03-16 18:03:09,551][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 18:03:09,553][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 18:03:09,553][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 18:03:09,553][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 18:04:13,897][transformers][INFO] - {'accuracy': 0.5909090909090909, 'RMSE': 28.06917861068948, 'QWK': 0.5302233902759528, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3776732460185698}
+[2025-03-16 18:04:13,897][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 18:04:13,899][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-608
+[2025-03-16 18:04:14,388][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 18:04:14,390][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 18:09:39,235][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-620
+[2025-03-16 18:09:39,701][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 18:09:39,703][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 18:09:39,859][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-608] due to args.save_total_limit
+[2025-03-16 18:09:39,871][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 18:09:39,873][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 18:09:39,873][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 18:09:39,873][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 18:10:44,058][transformers][INFO] - {'accuracy': 0.5909090909090909, 'RMSE': 28.06917861068948, 'QWK': 0.5302233902759528, 'HDIV': 0.007575757575757569, 'Macro F1': 0.3776732460185698}
+[2025-03-16 18:10:44,059][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 18:10:44,060][transformers.trainer][INFO] - Saving model checkpoint to /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-620
+[2025-03-16 18:10:44,398][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 18:10:44,400][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 18:10:44,605][transformers.trainer][INFO] - 
+
+Training completed. Do not forget to share your model on huggingface.co/models =)
+
+
+[2025-03-16 18:10:44,605][transformers.trainer][INFO] - Loading best model from /workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-576 (score: 0.579476861167002).
+[2025-03-16 18:10:44,655][transformers.trainer][INFO] - Deleting older checkpoint [/workspace/jbcs2025/outputs/2025-03-16/13-15-47/results/llama31_8b-balanced/C1/checkpoint-620] due to args.save_total_limit
+[2025-03-16 18:10:44,668][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 18:10:44,669][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 18:10:44,669][transformers.trainer][INFO] -   Num examples = 132
+[2025-03-16 18:10:44,669][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 18:11:49,145][transformers][INFO] - {'accuracy': 0.6287878787878788, 'RMSE': 26.285149626910837, 'QWK': 0.579476861167002, 'HDIV': 0.007575757575757569, 'Macro F1': 0.394676583276989}
+[2025-03-16 18:11:49,147][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 18:11:49,148][__main__][INFO] - Training completed successfully.
+[2025-03-16 18:11:49,148][__main__][INFO] - Running on Test
+[2025-03-16 18:11:49,148][transformers.trainer][INFO] - The following columns in the evaluation set don't have a corresponding argument in `PeftModelForSequenceClassification.forward` and have been ignored: essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text. If essay_year, grades, prompt, id_prompt, reference, essay_text, id, supporting_text are not expected by `PeftModelForSequenceClassification.forward`,  you can safely ignore this message.
+[2025-03-16 18:11:49,149][transformers.trainer][INFO] - 
+***** Running Evaluation *****
+[2025-03-16 18:11:49,149][transformers.trainer][INFO] -   Num examples = 138
+[2025-03-16 18:11:49,149][transformers.trainer][INFO] -   Batch size = 2
+[2025-03-16 18:12:58,833][transformers][INFO] - {'accuracy': 0.6594202898550725, 'RMSE': 25.931906372573962, 'QWK': 0.6043890865954924, 'HDIV': 0.007246376811594235, 'Macro F1': 0.43574433494234377}
+[2025-03-16 18:12:58,833][tensorboardX.summary][INFO] - Summary name eval/Macro F1 is illegal; using eval/Macro_F1 instead.
+[2025-03-16 18:12:58,834][transformers.trainer][INFO] - Saving model checkpoint to ./results/llama31_8b-balanced/C1/best_model
+[2025-03-16 18:12:59,177][transformers.configuration_utils][INFO] - loading configuration file config.json from cache at /root/.cache/huggingface/hub/models--meta-llama--Llama-3.1-8B/snapshots/d04e592bb4f6aa9cfee91e2e20afa771667e1d4b/config.json
+[2025-03-16 18:12:59,179][transformers.configuration_utils][INFO] - Model config LlamaConfig {
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 128000,
+  "eos_token_id": 128001,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "max_position_embeddings": 131072,
+  "mlp_bias": false,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 8,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": {
+    "factor": 8.0,
+    "high_freq_factor": 4.0,
+    "low_freq_factor": 1.0,
+    "original_max_position_embeddings": 8192,
+    "rope_type": "llama3"
+  },
+  "rope_theta": 500000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.49.0",
+  "use_cache": true,
+  "vocab_size": 128256
+}
+
+[2025-03-16 18:12:59,231][__main__][INFO] - Fine Tuning Finished.

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09e90d4b9670a78bba6dc5a3aa13e2057dac9e19edd2862943e50e1fd309c025
+size 5432