ChengManYu/qwen2-1.5b-ppt-outline-lora

Qwen2-1.5B PPT Outline Generator (LoRA Adapter)

This is the LoRA version, suitable for those who want to continue fine-tuning or conducting research.

Differences from the Merged version

• Merged：Ready to use, suitable for general users
• LoRA：Small in size and can continue to be fine-tuned on one's own data

How to use LoRA

from peft import AutoPeftModelForCausalLM
from transformers import AutoTokenizer

model = AutoPeftModelForCausalLM.from_pretrained(
    "ChengManYu/qwen2-1.5b-ppt-outline-lora",
    device_map="auto"
)

tokenizer = AutoTokenizer.from_pretrained("ChengManYu/qwen2-1.5b-ppt-outline-lora")

Training Information

• Use Qwen2-1.5B-Instruct as the base model
• LoRA rank = 16, alpha = 32
• Training for 3 epochs