DeepSeek v4.1 Flash EXL3 2.9 bpw for 2x DGX Sparks
Browser teleoperation data for embodied AI — elizaOS-ready episodes, trajectory_db converter, onchain provenance. The fuel station for agent robot stacks.
Running ONNX/TRT version of the MiniMax-H3 VAE in ComfyUI, which increase speed by up to 1.7x
北京大学未名超算队与北京大学学生 Linux 俱乐部合办的暑期 AI Infra 系列活动仓库
Privacy-aware trajectory viewer for local Codex task logs
Peer-metadata capture for chat-app calls (WhatsApp, Signal, Telegram ..etc). Authorized/consent use only. and more..
Video Delta Net hybrid attention for MiniMax-H3 as a native ComfyUI node.
Deployment kit: Qwen3.8-27B EXL3 3.5bpw target + DFlash2 EXL3 5.0bpw speculative draft — launcher, config, model cards
把 Codex imagegen 的工作流与提示词库带到任意智能体;支持 npm/pnpm 一键安装、配套 skill、独立 API 与 Qwen-Image 兼容网关。
Experimental Drosophila-inspired camera-to-robot bridge. Working mock neural backend, optical flow, IMU feedback and UDP motor control. MaleCNS integration planned.
پنل ساخت کانفیگ تکانموز
Autoregressive primitive-operation painter with GPU image-to-sequence converter
Heat Kernel Textures: the Geodesic Gaussians That Do Not Splat
CDAF (Cached Descriptive Asset Files) - open sidecar format for video so AI agents stop re-analyzing the same footage. Spec, CLI, agent skill, reproducible benchmark.
Finds the AI papers 28 tech companies put on arXiv over any date range, and splits lead authorship from bylines. A SKILL.md agent skill — no ML, no state, stdlib only.
Create camera-ready, editable academic figures from papers and method descriptions.
—
A small neural model for parsing date & time
We propose LeapTalk, a novel framework that achieves stable and real-time talking-head generation with a single forward step, scaling to arbitrarily long videos.
ComfyUI nodes and tested workflows for MiniMax H3 still-image generation, reference editing, and Qwen detail refinement.
Turn your OpenAI Dot into an OpenAI- and Claude-compatible API
Hermes Agent TTS and STT plugin for AllModels speech models and voices.
88 installable open-source Agent Skills for research, social intelligence, marketing, and business workflows—compatible with Codex, Claude Code, Cursor, Gemini CLI, and DeepSeek Harness.
Qoder Automation - Full Automation with UI Interaction (Auto-claim Pro Trial + 300 Credits)
Map the best way into any topic. 规划任何主题的最佳学习路径。
华为杯研究生数学建模竞赛全套AI-skills
小饼干英语视频工坊:从英语知识点到 30 秒竖屏教学视频的一站式 AI 创作工具。
A toolkit for exploring, gathering, crafting, and building in Minecraft with GPT-6 Astra computer use.
An agentic job search that runs on Claude Code: a recruiter, a crawler, and a market analyst writing to an Obsidian vault.
Canonical, machine-checkable record of the Justin Sun Prize: rules, awards, candidates, and verification evidence.
Hand-written NVFP4 W4A16 CUDA kernels for Volta
Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control
MerchantBench is a 365-day, order-level benchmark for evaluating the long-term coherence of LLM agents in seller-side e-commerce operations.
Slides Skill:让 AI Agent 生成可编辑 PPTD + PPTX,并附带本地浏览器编辑器 Slides skill for AI agents — generate editable PPTD + PPTX with a local browser editor
—
[EMNLP 2026] A multilingual benchmark for repository-level unit test generation, evaluating LLMs across 5 programming languages and real-world code under reproducible, Docker-based execution.
A library of experiments using frontier LLMs/VLMs for robotics tasks, currently focusing on GPT-6 Astra.
Type a movie name and get a fan edit: a captioned speech cut, then a montage cut to the beat.
—
—
—
OpenRouter alternatives: AI API aggregator and gateway options compared on model coverage, billing units, free tiers, OpenAI compatibility, rate limits and migration effort.
NVIDIA Sol-Attn for ComfyUI / Triton kernel on SM89 - SM121, with zero-copy MiniMax H3 nodes: memory-efficient attention, scheduled tau with graph preview, and feed-forward chunking. Measured 1.14–1.44× vs SageAttention and −37% MLP peak VRAM on H3
—
Build a person once, then shoot a whole series: framing, pose, placement, expression and aspect ratio vary while the person stays the same. ComfyUI nodes.
在终端(PowerShell / Windows Terminal / Linux Shell)中查看 A 股 K 线图、市盈率、成交量等基本信息的小工具。
End-to-end streaming data platform on AWS: Kafka → Iceberg lake → Kimball model → Athena, orchestrated with Airflow 3. Terraform, Glue PySpark, dbt.
—
Turn scattered notes, docs and transcripts into a queryable Markdown wiki — an LLM knowledge-base compiler with MCP access, no embeddings, self-hosted.
Benchmarking models as runtime Controllers for Loop Engineering
The open-source agentic framework to build, orchestrate, and operate production AI agents.
AI API aggregator companion: machine-readable catalog of 300+ models behind one key and one base URL, with routing patterns, cost accounting, OpenAI SDK and curl examples.
麦麦热量管家:减脂不是少吃你爱的,是先换掉不值这口热量的。基于麦当劳 MCP 的味热比点餐 Skill。
Terminal-first, knowledge-grounded multi-agent software delivery pipeline: scope requirements, implement changes, run tests, and gate pull requests with deterministic QA and ensemble code review.
—
Complete, end-to-end framework and high-performance OpenAI-compatible gateway for the Qoder ecosystem.
Code for SM4RT: Learning Structured Motion Geometry for 4D Reconstruction
—
Community registry of LLM serving-path traps that produce confidently wrong measurements: templates, tool parsers, reasoning fields, quant kernel paths, CUDA toolchains, KV allocation, eval harnesses, versioning. Symptom-first, with the check that catches each.
Unofficial TypeSafe Jev showcase — System One decisions, not chat.