|
--- |
|
base_model: Quasar-400B-X |
|
library_name: transformers |
|
model_name: Quasar-3.0-Max |
|
tags: |
|
- rl |
|
- silx |
|
- trl |
|
- sft |
|
licence: license |
|
--- |
|
|
|
# Quasar Series of Models |
|
|
|
<p align="center"> |
|
<img src="https://pbs.twimg.com/media/GnybON-W4AA3aRh?format=jpg&name=900x900" alt="Quasar Model Image" style="height: 500px;"> |
|
</p> |
|
|
|
## Introducing Quasar-3.0 |
|
|
|
This model is provided by **SILX INC**, Quasar-3.0-7B is a **distilled version** of the upcoming **400B Quasar 3.0** model. It is built upon the innovations introduced in the **Golden Formula in Reasoning** paper, featuring a novel training pipeline known as **TTM (Token Temperature Mechanism)** — a new approach to optimize reasoning and contextual focus during training. We also apply what we believe is the **best formula for Reinforcement Learning (RL) training** to date. |
|
|
|
### 🔥 Why Quasar-3.0 Matters |
|
|
|
This 7B model showcases the early strength and capability of the Quasar architecture. Despite its smaller size, it performs competitively and gives a glimpse of the power behind our full-scale 400B model. |
|
|
|
> We hope you put this model to good use and join us on the journey as we redefine reasoning in AI. |
|
|
|
Stay tuned for upcoming releases as we advance Quasar with full-scale RL enhancements and additional innovations. |
|
|
|
|
|
## Acknowledgements |
|
|
|
Special thanks to [Lambda](https://lambdalabs.com/) for their exceptional cloud computing platform that powered our training pipeline. Their GPU cloud infrastructure was instrumental in the development of this model. |
|
|
|
> "We couldn't have completed this training without Lambda's powerful computing resources. We highly recommend Lambda Cloud for machine learning and AI workloads." |
|
|
|
## About Lambda |
|
|
|
Lambda provides GPU cloud instances, on-demand GPU clusters, and GPU workstations specifically designed for machine learning and AI development. Their platform offers: |
|
|
|
- High-performance GPU instances |
|
- Cost-effective pricing |
|
- Easy scalability |
|
- Optimized ML/AI software environments |
|
|
|
Visit [Lambda's website](https://lambdalabs.com/) to learn more about their services and how they can accelerate your AI development. |
|
|
|
## Resources |
|
- [Research Paper](https://arxiv.org/abs/2412.06822) |
|
- [Website](https://sicopilot.cloud) |
|
|
|
## By |
|
- **SILX AI** |
|
- **Lambda Cloud** |
|
|
