Instructions to use JoshuaAshkinaze/argument-support with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use JoshuaAshkinaze/argument-support with PEFT:
from peft import PeftModel from transformers import AutoModelForSequenceClassification base_model = AutoModelForSequenceClassification.from_pretrained("answerdotai/ModernBERT-base") model = PeftModel.from_pretrained(base_model, "JoshuaAshkinaze/argument-support") - Notebooks
- Google Colab
- Kaggle
| { | |
| "base_model": "answerdotai/ModernBERT-base", | |
| "dataset": "ibm_debate_speeches", | |
| "target": "mostargumentssupport", | |
| "task": "regression", | |
| "num_labels": 1, | |
| "max_length": 1280, | |
| "best_config": { | |
| "learning_rate": 6e-05, | |
| "num_train_epochs": 12, | |
| "per_device_train_batch_size": 32, | |
| "gradient_accumulation_steps": 1, | |
| "lora_r": 128, | |
| "lora_alpha": 256, | |
| "lora_alpha_ratio": 2, | |
| "lora_dropout": 0.05, | |
| "target_modules": "Wqkv" | |
| }, | |
| "training_info": { | |
| "actual_epochs": 10.0, | |
| "best_epoch": 8.0, | |
| "best_metric": 0.7357637156286989, | |
| "best_model_checkpoint": "arg_qual/final_models/ibm_debate_speeches/mostargumentssupport/checkpoint-152" | |
| }, | |
| "test_metrics": { | |
| "test_test_loss": 0.3013608753681183, | |
| "test_test_spearman": 0.7058525835607535, | |
| "test_test_kendall_tau": 0.5102838725418962, | |
| "test_test_pearson": 0.6425724592223252, | |
| "test_test_rmse": 0.5489634825472827, | |
| "test_test_r2": 0.38694407752977733, | |
| "test_test_runtime": 5.3284, | |
| "test_test_samples_per_second": 24.397, | |
| "test_test_steps_per_second": 0.938 | |
| } | |
| } |