Faro-Qwen-4B / README.md
wenbopan's picture
Update README.md
e1a4a18 verified
---
license: mit
datasets:
- wenbopan/Fusang-v1
- wenbopan/OpenOrca-zh-20k
language:
- zh
- en
pipeline_tag: text-generation
---
![image/webp](https://cdn-uploads.huggingface.co/production/uploads/62cd3a3691d27e60db0698b0/s21sMRxRT56c5t4M15GBP.webp)
**The Faro chat model focuses on practicality and long-context modeling. It handles various downstream tasks with higher quality, delivering stable and reliable results even when inputs contain lengthy documents or complex instructions. Faro seamlessly works in both English and Chinese.**
# Faro-Qwen-4B
Faro-Qwen-4B is an improved [Qwen/Qwen1.5-4B-Chat](https://huggingface.co/Qwen/Qwen1.5-4B-Chat) with extensive instruction tuning on [Fusang-V1](https://huggingface.co/datasets/wenbopan/Fusang-v1). Compared to Qwen1.5-4B-Chat, Faro-Qwen-4B has gained greater capability in various downstream tasks and long-context modeling thanks to the large-scale synthetic data in Fusang-V1.
Faro-Qwen-4B uses dynamic NTK and continual training to extend its max context length to 100K. However, due to the lack of Dynamic NTK supports for`Qwen2ForCausalLM` in vLLM, inference on text longer than 32K requires using native `transformers` implementations.
## How to Use
Faro-Qwen-4B uses chatml template. I recommend using vLLM for long inputs.
```python
import io
import requests
from PyPDF2 import PdfReader
from vllm import LLM, SamplingParams
llm = LLM(model="wenbopan/Faro-Qwen-4B")
pdf_data = io.BytesIO(requests.get("https://arxiv.org/pdf/2303.08774.pdf").content)
document = "".join(page.extract_text() for page in PdfReader(pdf_data).pages) # 100 pages
question = f"{document}\n\nAccording to the paper, what is the parameter count of GPT-4?"
messages = [ {"role": "user", "content": question} ] # 83K tokens
prompt = llm.get_tokenizer().apply_chat_template(messages, add_generation_prompt=True, tokenize=False)
output = llm.generate(prompt, SamplingParams(temperature=0.8, max_tokens=500))
print(output[0].outputs[0].text)
```
<details> <summary>Or With Transformers</summary>
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained('wenbopan/Faro-Qwen-4B', device_map="cuda")
tokenizer = AutoTokenizer.from_pretrained('wenbopan/Faro-Qwen-4B')
messages = [
{"role": "system", "content": "You are a helpful assistant. Always answer with a short response."},
{"role": "user", "content": "Tell me what is Pythagorean theorem like you are a pirate."}
]
input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt").to(model.device)
generated_ids = model.generate(input_ids, max_new_tokens=512, temperature=0.5)
response = tokenizer.decode(generated_ids[0], skip_special_tokens=True)
```
</details>
For more info please refer to [wenbopan/Faro-Yi-9B](https://huggingface.co/wenbopan/Faro-Yi-9B)