---
base_model: Qwen/Qwen3-4B-Instruct-2507
datasets:
  - bio-nlp-umass/bioinstruct
library_name: peft
tags:
  - bio
  - bios-cli
---

# BIOS-CLI fine-tuned LoRA — `Qwen/Qwen3-4B-Instruct-2507` on `bio-nlp-umass/bioinstruct`

Trained via [BIOS CLI](https://github.com/bio-xyz/bios-cli).

## Run details

| field | value |
|---|---|
| base model | `Qwen/Qwen3-4B-Instruct-2507` |
| dataset | `bio-nlp-umass/bioinstruct` |
| examples | 500 |
| epochs | 3 |
| final loss | 37.9436 |
| checkpoint | `5e04f386-a16a-5245-ad4a-641870a94e06:train:0/sampler_weights/bios-sft-1778073817` |

## Inference

```python
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = AutoModelForCausalLM.from_pretrained('Qwen/Qwen3-4B-Instruct-2507')
tok = AutoTokenizer.from_pretrained('Qwen/Qwen3-4B-Instruct-2507')
# replace `<your-repo>` with this repo's id
model = PeftModel.from_pretrained(base, '<your-repo>')
```