Agentic AI 智能体开发实战
-
Updated
Jan 23, 2026 - Jupyter Notebook
Agentic AI 智能体开发实战
本地优先的 SFT 训练数据格式转换工具:将 JSON、JSONL、CSV 统一为 OpenAI messages,并生成 LLaMA-Factory 可用的 train.jsonl / val.jsonl。
网络安全领域大模型:Qwen3.8-27B LoRA 注入中文漏洞库知识,专家盲审 8.08/10(B 类 +83%)· vLLM+FP8 生产化 · 28 坑踩坑实录
An AI-powered technical support chatbot designed to help diagnose and resolve common software, infrastructure, and development issues. The assistant is built using a fine-tuned Mistral-7B model with LoRA adapters and deployed through a lightweight Gradio web interface. The system focuses on troubleshooting problems across multiple domains such as D
面向中文法律问答的 Qwen2.5-1.5B LoRA 微调:数据构建 → 双层引用评测 → GGUF 量化部署(单卡 T4 完整闭环)
LlamaFactory v0.9.4 WebUI Windows 打包器,一键启动大模型微调界面
Qwen3-0.6B 中文关键词抽取 LoRA 微调复现实验与 Base/Fine-tuned 对比
基于 LoRA 技术对大语言模型进行中文医疗问答领域的 SFT(监督微调)实验项目,并提供 reference-grounded 的 LLM-as-a-Judge 多维 pairwise 评测管线(并发调用、可审计证据、可视化报告)。
面向大模型指令微调的波音 737 FCOM 中文问答数据集 · 2765 条带思维链样本 · Alpaca/ShareGPT 格式 · 源文本取自 Eyderoe/FlightOperationsManual
LLaMA-Factory MM-Mix reference example for Online Dynamic Batching
Code, data pipeline, and evaluation tooling for PolyAlign: Conditional Human-Distribution Alignment.
Web-based LLM fine-tuning workbench that keeps training data in your own GPU environment.
To associate your repository with the llamafactory topic, visit your repo's landing page and select "manage topics."