Dialogue Summarization on SAMSum dataset.

Pipeline workflow -

Step 1: Parameter Efficient Fine-Tuning (PEFT) on FLAN-T5 base model.
Step 2: The results are then aligned using Reinforcement Learning from Human Feedback (RLHF).
Step 3: Retrieval Augmented Generation (RAG) to augment the generation of summaries.

Link to deployed application - https://huggingface.co/spaces/raghav-gaggar/Text_Summarizer