README.md

---
base_model: unsloth/Qwen2.5-3B-bnb-4bit
library_name: peft
pipeline_tag: text-generation
tags:
- lora
- qlora
- vietnamese
- lab21
license: apache-2.0
---

# Lab 21 — Qwen2.5-3B Vietnamese LoRA (r=16)

LoRA adapter fine-tuned on `5CD-AI/Vietnamese-alpaca-gpt4-gg-translated` (200 samples).

| Setting | Value |
|---------|-------|
| Base model | `unsloth/Qwen2.5-3B-bnb-4bit` |
| Rank / alpha | r=16, alpha=32 |
| Target modules | `q_proj`, `v_proj` |
| Eval perplexity | 4.55 |
| Student | Vũ Văn Học — 2A202600653 |

## Usage

```python
from peft import PeftModel
from unsloth import FastLanguageModel

base, tokenizer = FastLanguageModel.from_pretrained(
    "unsloth/Qwen2.5-3B-bnb-4bit", load_in_4bit=True
)
model = PeftModel.from_pretrained(base, "2vhoc/lab21")
```