A ChatBot based on the ChatGLM LoRA Finetuning Project : https://github.com/lich99/ChatGLM-finetune-LoRA/
We fine-tuned the LLM with LoRA and got a TalkMate that speaks like a REAL Human!
The Chat Example using the backbone model and fine-tuned model is here: https://github.com/ZeyuBa/ChatGLM-6B_LoRA_TalkMate/tree/main/chat_example
BlEU Score and Train loss is:
| epoch | BLEU score |
|---|---|
| - | 0.478 |
| 0 | 0.545 |
| 1 | 0.540 |
| 2 | 0.529 |
