| --- |
| license: apache-2.0 |
| library_name: transformers |
| --- |
| # Helios-Reasoner-7B |
| <!-- markdownlint-disable first-line-h1 --> |
| <!-- markdownlint-disable html --> |
| <!-- markdownlint-disable no-duplicate-header --> |
|
|
| <div align="center"> |
| <img src="figures/fig1.png" width="60%" alt="Helios-Reasoner-7B" /> |
| </div> |
| <hr> |
|
|
| <div align="center" style="line-height: 1;"> |
| <a href="LICENSE" style="margin: 2px;"> |
| <img alt="License" src="figures/fig2.png" style="display: inline-block; vertical-align: middle;"/> |
| </a> |
| </div> |
| |
| ## 1. Introduction |
|
|
| Helios-Reasoner-7B is our flagship open reasoning model. Through extended RLHF and a curated math/code post-training mix, it now closes much of the gap with frontier closed models on hard multi-step reasoning. |
|
|
| <p align="center"> |
| <img width="80%" src="figures/fig3.png"> |
| </p> |
|
|
| ## 2. Evaluation Results |
|
|
| ### Comprehensive Benchmark Results |
|
|
| <div align="center"> |
|
|
| | | Benchmark | Falcon-S | Mistral-Lite | Helios-v1 | Helios-Reasoner-7B | |
| |---|---|---|---|---|---| |
| | **Core Reasoning Tasks** | Math Reasoning | 0.533 | 0.515 | 0.519 | 0.573 | |
| | | Logical Reasoning | 0.808 | 0.781 | 0.830 | 0.838 | |
| | | Common Sense | 0.699 | 0.723 | 0.714 | 0.750 | |
| | **Language Understanding** | Reading Comprehension | 0.684 | 0.679 | 0.669 | 0.718 | |
| | | Question Answering | 0.606 | 0.571 | 0.584 | 0.618 | |
| | | Text Classification | 0.791 | 0.797 | 0.818 | 0.838 | |
| | | Sentiment Analysis | 0.767 | 0.774 | 0.755 | 0.800 | |
| | **Generation Tasks** | Code Generation | 0.641 | 0.651 | 0.668 | 0.674 | |
| | | Creative Writing | 0.593 | 0.624 | 0.603 | 0.636 | |
| | | Dialogue Generation | 0.617 | 0.613 | 0.632 | 0.660 | |
| | | Summarization | 0.728 | 0.736 | 0.739 | 0.779 | |
| | **Specialized Capabilities** | Translation | 0.780 | 0.764 | 0.783 | 0.811 | |
| | | Knowledge Retrieval | 0.637 | 0.677 | 0.661 | 0.688 | |
| | | Instruction Following | 0.755 | 0.751 | 0.719 | 0.770 | |
| | | Safety Evaluation | 0.693 | 0.702 | 0.709 | 0.750 | |
|
|
| </div> |
|
|
| ### Overall Performance Summary |
| The Helios-Reasoner-7B demonstrates strong performance across all evaluated benchmark categories, with particularly notable results in reasoning and generation tasks. |
|
|
| ## 3. Chat Website & API Platform |
| We offer a chat interface and API for you to interact with Helios-Reasoner-7B. Please check our official website for more details. |
|
|
| ## 4. How to Run Locally |
|
|
| Please refer to our code repository for more information about running Helios-Reasoner-7B locally. |
|
|
| ### Temperature |
| We recommend setting the temperature parameter to 0.6. |
|
|
| ## 5. License |
| This repository is released under the apache-2.0 license. The model supports commercial use. |
|
|
| ## 6. Contact |
| If you have any questions, please contact us at research@helios-labs.ai. |
|
|
