Back to directory
AI & Automationsongrui111/dsh-fine-tune
dsh-fine-tune
Local LoRA / QLoRA fine-tuning for DeepSeek Harness: import JSONL, train on your own GPU (or a remote SSH host), evaluate, and export an adapter. One-click environment install; Windows + NVIDIA verified. 本地微调插件:数据集→训练→评测,支持本机与远程 GPU。
Suggested install command
npm view dsh-fine-tuneAlways inspect the linked repository and skill instructions before running commands. Skills are instructions; permissions and execution still matter.
Compatibility
Agent support matrix
3 supported
| Agent | Status |
|---|---|
| Claude Code | Supported |
| OpenCode | Not listed |
| Cursor | Supported |
| MCP | Not listed |
| GitHub Copilot | Not listed |
| Windsurf |