fredy-zhang/prdit
[ICLR 2026] Official Implementation of paper "Pixel-Level Residual Diffusion Transformer: Scalable 3D CT Volume Generation"
GITHUB SIGNALS / PROJECT INDEX
Search the archive. Follow your curiosity. Keep the projects you want to come back to.
[ICLR 2026] Official Implementation of paper "Pixel-Level Residual Diffusion Transformer: Scalable 3D CT Volume Generation"
This is the repository for “Agent-guided Reliability-aware Temporal Mask Evolution for Imperfectly Supervised Video Polyp Segmentation”
End-to-end LLM training based on Gemma 3.
Who the good traders on Robinhood Chain are buying: resolved wallets, a 20-second on-chain tape, provenance on every fill, AI verdicts, bursts and exits. Site + Telegram bot + API. Runs on $0/month.
Fish-style zsh history autosuggestions ranked by Jev (TypeSafe)
A self-hosted drop-in replacement for TypeSafe's jev, powered by GliFormer.
Open the project to explore the idea and the code.
An autonomous red-teaming engine for LLMs. RedThread manages the full security lifecycle: generating adversarial attacks, executing precision evaluations, and synthesizing validated guardrails for safe self-improvement.
A recreation of Ragnarok Online client in Go with gogpu
Clarity - an Agent skill for clearer writing.
Self-hosted AI workspace with agents, skills, and tools (Gmail, Calendar) that runs entirely on your own provider API keys (BYOK). Bring your own keys — Groq, OpenRouter, NVIDIA, Hugging Face, Google AI.
Local-first MCP gateway. One port for every tool and every AI client: lazy discovery (~90% token savings), tool integrity + quarantine, secrets in the OS keychain.
⚡ Drop Android Virtual Device (AVD) emulator RAM from ~8GB to ~1.5GB on Apple Silicon & Linux. Inspired by simslim.
Interactive 3D explorer of a desktop computer, from an assembled ATX tower down to a GPU streaming multiprocessor.
LLM inference in C/C++.
Open the project to explore the idea and the code.
A local inference engine for Apple silicon, built around the model.
Qwen3.8-Flash-Next on 2× RTX 3090 + 128 GB RAM. Experimental peaks: 1,860 tok/s prefill at 260K input; 89.1 tok/s decode at 128K input. Full 256K context.
Your app dock. Extendable with your own plugins. Linux, Windows, macOS (beta).
Are you running too few agents, or too many? AgentPlayback visualizes your coding agents so you can see when each one ran, when it was blocked on you, and what it cost.
Detect youtube sponsor segment with live audio and transcript powered by Jev
Open, Jev-compatible System One decision server on DiffusionGemma
Runtime behavioral ablation for compressed LLMs. First target: Ternary Bonsai 2 27B — no weight modification or re-quantization. by OrcaRouter team
🎭 AI 写小说:从零生成 10-50 章完整中文小说,三层问答 · 创作记忆 · 悬念钩子 · 自动校验,长篇网文连载皆宜|开源免费,适配主流 coding agent|AI novel writing skill