OpenHands LM 32B AWQ
| Property | Value |
|---|---|
| Parameter Count | 32 Billion |
| Quantization | INT4 GEMM (AutoAWQ) |
| Context Window | 128K tokens |
| Model URL | https://huggingface.co/stelterlab/openhands-lm-32b-v0.1-AWQ |
What is openhands-lm-32b-v0.1-AWQ?
OpenHands LM is an open-source language model specifically designed for software development tasks. Built on Qwen Coder 2.5 Instruct 32B, this model has been fine-tuned using an RL-based framework and quantized to INT4 for efficient deployment. The model achieves impressive performance metrics, matching the capabilities of much larger models while maintaining accessibility for local deployment.
Implementation Details
The model utilizes AutoAWQ quantization technology to compress the original weights while maintaining performance. It features a substantial 128K token context window, making it suitable for processing large codebases and extended software engineering tasks.
- Built on Qwen Coder 2.5 Instruct 32B foundation
- Trained using RL-based framework from SWE-Gym
- INT4 quantization for efficient deployment
- 128K token context window
Core Capabilities
- 37.2% verified resolve rate on SWE-Bench Verified
- Performance comparable to 671B parameter models
- Specialized in GitHub issue resolution
- Local deployment capability on consumer hardware
Frequently Asked Questions
Q: What makes this model unique?
This model achieves remarkable efficiency, matching the performance of models 20x larger while being deployable on consumer hardware like a single 3090 GPU. Its INT4 quantization makes it particularly accessible for local deployment.
Q: What are the recommended use cases?
The model excels at solving GitHub issues and software engineering tasks. However, it's currently optimized for issue resolution and may perform less consistently on more varied software engineering tasks. Users should note that it may sometimes generate repetitive steps and is sensitive to quantization levels.
