Long Horizon Terminal Benchmark with Dense Reward Grading
-
Updated
Jul 30, 2026 - Python
Long Horizon Terminal Benchmark with Dense Reward Grading
Tencent Hy3 295B MoE (NVFP4) on 2x NVIDIA DGX Spark — TP2 over 200GbE, 256K context, 26 tok/s end-to-end. Tuned, benchmarked, agent-ready.
First public recipe for Tencent Hy3-295B (295B MoE, NVFP4) across two NVIDIA DGX Spark GB10 nodes: vLLM TP=2, MTP speculative decoding, and the full toolchain-debugging story including what fails.
Hy3-powered microscopy, FLIM, and scientific RAG research agent
Read-only Hy3 spec-to-diff reviewer with a loopback Web UI and CLI, local citation validation, and Markdown/JSON PR-readiness reports via TokenHub.
Production-hardened OpenAI-compatible API proxy for Tencent Hy3 (295B MoE) via HuggingFace Gradio API. 25 bug fixes over the original — Dockerfile fix, CORS compliance, admin auth, input size limits, concurrency limiter fixes, and more.
Issue-killer: Hy3-powered GitHub Issue contribution planner for Tencent Rhino-Bird 2026
HY3 Desktop — Tencent Hunyuan 3 assistant that also codes, runs agents & builds Excel/decks. One-click, free 1 month. Apache 2.0. Win/Mac.
Hy3-powered local context relay for AI coding tools
Add a description, image, and links to the hy3 topic page so that developers can more easily learn about it.
To associate your repository with the hy3 topic, visit your repo's landing page and select "manage topics."