《开源大模型食用指南》针对中国宝宝量身打造的基于Linux环境快速微调（全参数/Lora）、部署国内外开源大模型（LLM）/多模态大模型（MLLM）教程

A Gemini 2.5 Flash Level MLLM for Vision, Speech, and Full-Duplex Multimodal Live Streaming on Your Phone

VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning

Firefly: 大模型训练工具，支持训练Qwen2.5、Qwen2、Yi1.5、Phi-3、Llama3、Gemma、MiniCPM、Yi、Deepseek、Orion、Xverse、Mixtral-8x7B、Zephyr、Mistral、Baichuan2、Llma2、Llama、Qwen、Baichuan、ChatGLM2、InternLM、Ziya2、Vicuna、Bloom等大模型

Easy-to-Use RAG Framework; CCF AIOps International Challenge 2024 Top3 Solution; CCF AIOps 国际挑战赛 2024 季军方案

Inferflow is an efficient and highly configurable inference engine for large language models (LLMs).

Explore LLM model deployment based on AXera's AI chips

A custom ComfyUI node for MiniCPM vision-language models, supporting v4, v4.5, and v4 GGUF formats, enabling high-quality image captioning and visual analysis.

Advanced Android photo editor powered by on-device AI models (LaMa, EdgeSAM, MobileViT) with GPU-accelerated adjustments, voice commands, generative fill, and professional editing tools. Built with Jetpack Compose and ONNX Runtime for fast, offline-first image processing.

CRNN/LPRNet/STNet + CTC + CCPD

MaskClaw项目构建了一套基于端侧多模态大模型的自进化隐私防护架构，通过视觉脱敏与Agentic 规则演进，实现了 Agent 行为偏好与隐私防护策略的动态协同生长，确保了高危隐私数据在云端交互中的物理级安全与个体化适配。

PicQ: Demo for MiniCPM-o 2.6 to answer questions about images using natural language.

Build voice apps fast. Unified API for speech recognition & synthesis with streaming, WebSocket, and multi-engine support.

The proof is in the pudding. Real-time AI video commentary with text-to-speech, running locally on your GPU.

AI Chat, Video feed, and Picture Generation

ajemnminecraft@gmail.com

Build local voice AI solutions with Vox, offering speech-to-text, text-to-speech, and voice chat—all without cloud dependence or API keys.

🧪 Testing OpenBMB's on-device AI models (MiniCPM series) with Dayflow. Forked from JerryZLiu/Dayflow

An innovative anomaly detection system for manufacturing industries. Integrating MVTec AD dataset, MiniCPM language model, and LoRA explanation generation, achieving AUC 0.7175 practical-level anomaly detection performance on 416 complete wallplugs dataset.

Grounded Semantic Role Labeling for Human–Robot Interaction — synthetic multimodal data generation and MiniCPM-V training for grounded command understanding. (EMNLP 2025)