library_name: transformers
datasets:
- codeparrot/apps
- BAAI/TACO
- AI-MO/NuminaMath-CoT
language:
- en
base_model:
- Qwen/Qwen2.5-32B-Instruct
Model Details
Model Description
This is a 32B reasoning model trained from Qwen2.5-32B-Instruct with 17K data. The performance is on par with o1-preview model on both math and coding. Please see our blog post for more details.
- Developed by: NovaSky Team from Sky Computing Lab at UC Berkeley.
Training Details
Training Data
17K verified correct responses from Qwen/QwQ-32B-Preview on coding, math. In addition, we add the science portion from the Still-2 paper.
Training Procedure
We perform supervised fine tuning on the data, with a batch size of 96.
Speeds
We use Llama-Factory for training. On 8 H100, the training takes 19 hours with DeepSpeed Zero-3 Offload.
Evaluation
| Model | Math500 | AIME2024 | LiveCodeBench-Easy | LiveCodeBench-Medium | LiveCodeBench-Hard | GPQA-Diamond |
|---|---|---|---|---|---|---|
| Qwen-2.5-3 2B-Instruct | 85.2 | 16.7 | 82.4 | 40.0 | 8.9 | 42.9 |
| Sky-T1 | 88.6 | 43.3 | 87.9 | 54.4 | 17.1 | 53.5 |
| QwQ | 90.6 | 50.0 | 88.7 | 57.3 | 17.9 | 56.6 |
| o1-preview | 85.5 | 46.6 | 92.0 | 56.6 | 13.8 | 73.3 |
Acknowledgement
We would like to thanks the compute resources from Lambda Lab and AnyScale. We would like to thanks the academic feedback and support from the Still-2 Team, and Junyang Lin from the Qwen Team.
Citation
Please considering citing our blog post if you found it useful for your research. Thank you!
@misc{sky_t1_2025,
author = {NovaSky Team},
title = {Sky-T1: Fully open-source reasoning model with o1-preview performance in $450 budget},
howpublished = {https://novasky-ai.github.io/posts/sky-t1},
note = {Accessed: 2025-01-09},
year = {2025}
}