|
|
---
|
|
|
license: apache-2.0
|
|
|
---
|
|
|
|
|
|
# QwQ-Math-1.5B-Persona
|
|
|
|
|
|
## Introduction
|
|
|
|
|
|
QwQ-Math-1.5B-Persona is finetuned from Qwen2.5-Math-1.5B-Instruct on 1 million math persona data (see [this paper](https://arxiv.org/abs/2406.20094) for details about how to construct the data).
|
|
|
|
|
|
Currently QwQ-Math-1.5B-Persona is meant to serve as a draft model for losslessly accelerating the inference of QwQ-32B, but you may also use it as a standalone model.
|
|
|
|
|
|
## Quickstart
|
|
|
|
|
|
Here is a code snippet for using QwQ-Math-1.5B-Persona to accelerate the inference of QwQ 32B:
|
|
|
|
|
|
```python
|
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
|
"Qwen/QwQ-32B-Preview",
|
|
|
torch_dtype="auto",
|
|
|
device_map={'': 0}
|
|
|
)
|
|
|
|
|
|
draft_model = AutoModelForCausalLM.from_pretrained(
|
|
|
"Geralt-Targaryen/QwQ-Math-1.5B-Persona",
|
|
|
torch_dtype="auto",
|
|
|
device_map={'': 0}
|
|
|
)
|
|
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained("Qwen/QwQ-32B-Preview")
|
|
|
|
|
|
prompt = "How many r in strawberry."
|
|
|
messages = [
|
|
|
{"role": "system", "content": "You are a helpful and harmless assistant. You are Qwen developed by Alibaba. You should think step-by-step."},
|
|
|
{"role": "user", "content": prompt}
|
|
|
]
|
|
|
text = tokenizer.apply_chat_template(
|
|
|
messages,
|
|
|
tokenize=False,
|
|
|
add_generation_prompt=True
|
|
|
)
|
|
|
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
|
|
|
|
|
generated_ids = model.generate(
|
|
|
**model_inputs,
|
|
|
max_new_tokens=512,
|
|
|
assistant_model=draft_model
|
|
|
)
|
|
|
generated_ids = [
|
|
|
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
|
|
]
|
|
|
|
|
|
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
|
|
```
|
|
|
|
|
|
For the more advanced SVIP draft length policy, please refer to [this GitHub repo](https://github.com/Geralt-Targaryen/SVIP).
|
|
|
|
|
|
## Citation
|
|
|
|
|
|
If you find QwQ-Math-1.5B-Persona to be helpful, please cite the following paper.
|
|
|
|
|
|
```
|
|
|
@misc{zhang2024svip,
|
|
|
title={Draft Model Knows When to Stop: A Self-Verification Length Policy for Speculative Decoding},
|
|
|
author={Ziyin Zhang and Jiahao Xu and Tian Liang and Xingyu Chen and Zhiwei He and Rui Wang and Zhaopeng Tu},
|
|
|
year={2024},
|
|
|
eprint={2411.18462},
|
|
|
archivePrefix={arXiv},
|
|
|
primaryClass={cs.CL},
|
|
|
url={https://arxiv.org/abs/2411.18462},
|
|
|
}
|
|
|
```
|
|
|
|
