9.3k repos tracked
updated 2d ago
New

LLM & Inference

Models, serving stacks, quantisation and fine-tuning.

Trending New Most starred 131 repositories · topics: llm, large-language-models, llm-inference, llm-serving, fine-tuning
  1. 101
    A design system for AI applications. Material, physical, tactile, technical.
    TypeScript created 26d ago pushed 7d ago agent-uiaiai-agents
    +1.0/wk
    measured
    ★ 2k
    25
  2. 102
    DeepSeek-V4.1-Flash-1017 EXL3 2.9bpw on 2x NVIDIA DGX Spark (GB10) with TensorFold: 2.1-2.3x the vLLM kit's decode speed.
    Python created 21d ago pushed 3d ago deepseekdgx-sparkexl3
    +5.0/wk
    measured
    ★ 76
    25
  3. 103
    Local AI desktop app: a Chromium browser with a GPU-aware llama.cpp runtime, OpenAI-compatible API, AI chat and agents, Stable Diffusion image/video, TTS and music, plus an MCP server so Claude Code, Codex and Cline can drive your browser. Windows, macOS, Linux.
    JavaScript created 29d ago pushed 3d ago ai-agentsai-browserbrowser-automation
    +4.0/wk
    measured
    ★ 132
    25
  4. 104
    👋 Hello Agent System — learn enterprise AI agent system design by building every production layer from scratch. 17 bilingual lessons (EN/中文) with tested exercises: tools, architectures, reliability, security, evals, distributed execution, cost, RAG, release ops. 企业级 Agent 系统设计训练营
    Python created 42d ago pushed 40d ago agentai-agentsbilingual
    +1.0/day
    measured
    ★ 312
    25
  5. 105
    精酿 BrewReel:让 DeepSeek 这类便宜模型也能做出好看的竖版宣传片。写一份产品简报,AI 挑镜头、写文案,一条命令出片。3 种配方、6 个行业、广告法校验,开源可商用。
    JavaScript created 43d ago pushed 27d ago agent-skillsclaude-codedeepseek
    +5.0/wk
    measured
    ★ 180
    24
  6. 106
    The AI chat app that sweats the details. Every top model, real memory, living documents, and your own tools.
    TypeScript created 36d ago pushed 24d ago aiai-chatchat
    +2.0/wk
    measured
    ★ 512
    24
  7. 107
    clodfarm (say it out loud): a farm of Claude Code agents. Plant a mission, they split it into sub-agents, open the work, and pace themselves on each account's real 5-hour and weekly usage. Steer it from the Claude app.
    Python created 44d ago pushed 16d ago agent-orchestrationaiai-agents
    +3.0/wk
    measured
    ★ 217
    24
  8. 108
    Sno Station Skills: 31 tested skills that turn Claude Code and Codex into one working team. Cross-vendor peer review, automatic handoff when an agent hits its quota, a one-page morning brief, work orders with proof, and a nightly loop that learns from your own sessions. Open source, installed with one sno setup.
    Shell created 38d ago pushed 3d ago agent-skillsagentic-codingai-code-review
    −3.3/day
    measured
    ★ 861
    23
  9. 109
    TOMC Memory: CPU context compression and task memory for Codex, Claude Code, Cursor, TRAE, WorkBuddy and other MCP clients.
    Python created 32d ago pushed 3d ago agent-memoryclaude-codecodex
    +1.0/wk
    measured
    ★ 267
    23
  10. 110
    CodeAgent is an open-source coding agent that lives in your terminal. It understands your repository, plans multi-step changes, writes and edits files, runs commands and tests, and iterates until the task is done — all with a transparent, extensible tool loop.
    Python created 37d ago pushed 25d ago agentai-agentcli
    +1.0/wk
    measured
    ★ 1k
    23
  11. 111
    MCP desktop companion for AI agent status, usage tracking, and everyday tools. Anything Universe · 桌面伙伴与日常工具入口。
    Python created 28d ago pushed 10d ago ai-agentanything-universedesktop-pet
    +1.0/wk
    measured
    ★ 313
    23
  12. 112
    A multi-model Muse: personal AI voice assistant on ESP32 - wake word or knob, any OpenAI-compatible brain (STT/chat/TTS mix-and-match), streaming replies, out of the box. Built on muse-gadget-sdk.
    Shell created 26d ago pushed 19d ago ai-assistantdeepseekedge-ai
    +2.0/wk
    measured
    ★ 262
    23
  13. 113
    Component layer for AI agents: implementation-aware contracts for building project-native interface components.
    JavaScript created 44d ago pushed 3d ago agent-skillsai-agentsclaude-code
    +1.0/wk
    measured
    ★ 192
    23
  14. 114
    ARTEX · AI autonomous penetration testing system (multi-agent LLM, Go + Next.js) · AI 自主渗透测试系统 · 百度 agent+ 攻防挑战赛冠军
    Go created 30d ago pushed 20d ago ai-agentai-securityartex
    −123/day
    measured
    ★ 1.3k
    23
  15. 115
    Build AI phone agents on your own computer and test every change with an AI caller. Open source, runs free with no API keys, cents per call with your own keys. Imports Dialogflow CX.
    Python created 20d ago pushed 6d ago agents-as-codecontact-centerconversational-ai
    −5.7/day
    measured
    ★ 507
    23
  16. 116
    AI assistant and MCP server for Blender: visual chat, scene tools, multi-agent workflows. Works with OpenAI, ChatGPT, Claude, Gemini, Ollama, OpenCode. Blender AI addon and plugin, copilot for 3D, bpy Python scripting, MCP integration.
    Python created 23d ago pushed 17d ago ai-assistantblenderblender-addon
    —
    stale
    ★ 976
    22
  17. 117
    麦麦省钱助手 —— 基于麦当劳 MCP 的点餐成本优化 Skill。算出当前门店最省钱的点餐方案,并主动提示凑单陷阱。
    Python created 28d ago pushed 17d ago agentcost-optimizationcoupon
    −1.0/wk
    measured
    ★ 992
    22
  18. 118
    🌃 Capri — Edgerunners-inspired AI agent for your desktop. Works with Claude Code, Codex, Antigravity & more.
    created 11d ago pushed 9d ago ai-agentai-agentsai-assistants
    —
    stale
    ★ 386
    22
  19. 119
    🍟 麦当劳中国官方 MCP 智能助手技能:把 35 个原子工具编排成 4 条业务流 —— 活动推送 · 联名关注 · 最优惠点餐 · 多人群点聚合账单。算价用 calculate-price 真算,下单必二次确认。|McDonald's China MCP agent skill: campaigns, collab watch, best-value ordering, group bill splitting.
    created 30d ago pushed 11d ago agent-skillsaiai-agent
    −2.0/wk
    measured
    ★ 379
    22
  20. 120
    构屎 GOUJI — 纯 C++/CUDA 的三值 mini LLM。2 bit/权重,零 Python;联想记忆工具路由,零训练参数;143 个缺陷与 29 个变异测试记录在案。
    C++ created 21d ago pushed 20d ago associative-memorycppcuda
    +2.0/wk
    measured
    ★ 97
    21
  21. 121
    Unofficial skill that teaches coding agents to build with TypeSafe AI's Jev: typed decisions, calibrated confidence, and prior art from 150+ community projects.
    Python created 43d ago pushed 37d ago agent-skillagent-skillsai-agents
    +2.0/wk
    measured
    ★ 150
    21
  22. 122
    AI novel generation with world building, interactive choices, and persistent character state. Next.js + Go. AI 小说创作台。
    Go created 36d ago pushed 36d ago aiai-novelgolang
    +1.0/wk
    measured
    ★ 236
    20
  23. 123
    Self-hosted decision model (selfjev-4b) with a Jev-compatible API: Python SDK, server, fine-tuning and RLCD.
    Python created 17d ago pushed 17d ago classificationdecisionsfine-tuning
    +1.0/wk
    measured
    ★ 60
    20
  24. 124
    STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization
    Python created 39d ago pushed 16d ago large-language-modelsllmpost-training-quantization
    flat
    measured
    ★ 101
    20
  25. 125
    A Markdown convention and linter for project docs written by AI and read by humans and agents
    Rust created 41d ago pushed 26d ago aiai-agentsclaude-code
    −1.0/wk
    measured
    ★ 194
    20
  26. 126
    Blender add-on: create and edit 3D models by chatting with AI - send text or an image, get working bpy code
    Python created 30d ago pushed 29d ago 3d-graphicsaiblender
    —
    stale
    ★ 218
    20
  27. 127
    Switchless RoCE collectives, NCCL profiles and fabric tooling for DGX Spark inference recipes
    Python created 34d ago pushed 6d ago connectx-7cudadgx-spark
    flat
    measured
    ★ 41
    19
  28. 128
    辞达:用你选择的大模型,在 Mac 的任何地方翻译和润色文字。Translate and polish text anywhere on your Mac.
    Swift created 44d ago pushed 31d ago llmmacosmenubar-app
    —
    stale
    ★ 173
    19
  29. 129
    Research and writing can be automated. Authorship cannot. Evidence list + agent skills for spotting AI-produced papers no human stood behind.
    Python created 34d ago pushed 32d ago agent-skillsai-generated-content-detectionclaude-code
    flat
    measured
    ★ 146
    19
  30. 130
    A local app that rents a GPU in your own Colab account and serves Qwen3.8 on it: 27B on an A100-40G, or Flash-Next (125B-A6B MoE) on an A100-80G. The cost is shown before it starts; chat with images; Codex or any OpenAI client uses a fixed local /v1. Auto-stop, CU ledger, live prefill/decode. Windows, macOS, Linux.
    Python created 44d ago pushed 28d ago colabexllamav3google-colab
    flat
    measured
    ★ 103
    19
  31. 131
    Watches company job boards, scores openings against your resume, and emails you when something worth applying to shows up.
    Python created 44d ago pushed 38d ago automationjob-searchllm
    —
    stale
    ★ 193
    19
← previous page 3 of 3