This repository contains the final Assignment 3 model artifacts for the QLoRA path (Part C) and downstream PatentSBERTa fine-tuning (Part D).
comparison_table.csv: Required Part E comparison of Baseline vs Assignment 2 vs Assignment 3 F1 scores.
config.json: Model architecture and label configuration for the final sequence-classification model.
model.safetensors: Trained model weights from Assignment 3 Part D (the final model to evaluate/use).
part_d_summary.json: Compact summary of Part D setup, dataset sizes, and final performance metrics.
special_tokens_map.json: Mapping of tokenizer special tokens used with the model.
tokenizer_config.json: Tokenizer runtime settings for preprocessing text consistently.
tokenizer.json: Full tokenizer definition used during training/inference.
training_args.bin: Serialized Hugging Face training arguments for reproducibility of Part D.
vocab.txt: Token vocabulary file required by the tokenizer.
Together, these files document and reproduce the final…
