GitHub Repo

GitHub 熱門專案、開源工具、框架、runtime 與重要 release。

Edge0 Streams MoE Experts from SSD, Requiring Just 2.9 GiB of Active Memory for a 35B Model at Short Context Lengths

The new open-source framework uses predictive routing to preload experts for the next step and Recover-LoRA to compensate for int4 quantization loss, so the entire 23GB checkpoint does not need to remain resident in memory. Official M4 Pro tests measured 14.9–17.7 tokens per second, but the framework currently supports only Apple Silicon, and its memory and quality results have yet to be independently reproduced.

2026-09-11

推論系統

NVIDIA Dynamo Splits Multimodal Inference into Three Stages, Cutting Time to First Token by Up to 42.2% Under Mixed Traffic

Dynamo can now assign visual encoding, LLM prefill, and decoding to separate workers and different GPU tiers, preventing image requests from blocking text-only traffic. Although NVIDIA’s tests show improvements of up to 5× in time to first token and 7× in end-to-end latency, long outputs or large dense models may offset the benefits of disaggregation.

2026-09-11

推論系統

Delaying the Submission of Ready Turns in Agentic Workflow Scheduling Cuts P95 Completion Time by Up to 71.4% Under Congestion

New research separates a turn becoming ready from its submission to the inference engine, using CVaR tail risk and a dynamic workload budget to determine submission order. SWE-bench trace replays achieved up to a 3.5× P95 speedup, but the results cover only fixed agent trajectories, a single randomized arrival seed, and an A100/vLLM environment.

2026-09-11

AI 安全

RAG-Safety-Bench Isolates Retrieval Quality as a Variable, Showing Safety Fine-Tuning May Not Withstand Answer-Bearing Documents

The new benchmark holds retrieved content constant and uses four conditions to measure how RAG changes five open models’ responses to harmful questions. Documents containing answers made every model more likely to produce accurate but dangerous content, while safety judges showed only moderate agreement.

2026-09-11

具身 AI 與機器人

Show-Harness Controls Different Robots with Discrete Semantic Actions, Achieving an 89% Success Rate Across Ten Tasks

Show-Harness has a general-purpose VLM output symbols for directions, grasping, and releasing, which robot-specific interpreters then translate into constrained motions. Gemini in a zero-shot setting and a fine-tuned Qwen 2B achieved success rates of 89% and 86%, respectively, across ten real-world tasks, though testing remains focused on tabletop pick-and-place.

2026-09-11

推論系統

PELM Jointly Tunes DVFS and Speculative Decoding, Cutting On-Device LLM Energy Consumption by Up to 52.4%

PELM goes beyond adjusting processor frequency: it dynamically selects speculative decoding configurations and verification depth, jointly controlling hardware and model workload under thermal constraints. The paper reports speedups of up to 23.1% and energy savings of up to 52.4%, but these peak results cannot be directly extrapolated to smartphones or other models.

2026-09-11

地球觀測與應用研究

MAPL-EMIT Trains Satellite Methane Detection on 3.6 Million Simulated Plumes, Releases Model and Global Dataset

Google and NASA JPL enable a vision Transformer to directly combine EMIT’s spectral and spatial context, simultaneously estimating concentrations, plume boundaries, and emission-source locations. The publicly released model recovers about 80% of expert-labeled cases, but its higher sensitivity also introduces risks of surface confusion and false positives.

2026-09-11

生成式音訊

YuE2-3B Plans Full Songs with Editable Scores, but Only Its Best-of-8 Results Surpass Suno v5

YuE2 combines symbolic melody and chord planning, semantic tokens, and flow-matching audio synthesis in a single generation pipeline. Its weights can run on a 24GB GPU, but its noncommercial license and best-of-8 evaluation mean comparisons with commercial systems should still be interpreted cautiously.

2026-09-11

代理工具

AWS Open-Sources Pizza Bot, Turning Long-Running Agent Work into a Resumable Asynchronous Inbox

Pizza Bot stores agent state, approval requests, and scheduled results on the server, allowing work to continue after users leave the conversation. It supports multiple model providers, MCP, and Agent Skills, but local-first does not mean tool-safe: remote deployments and extensions still require carefully designed trust boundaries.

2026-09-11

應用研究

NASA and IBM Open-Source Lunar Foundation Model, Unifying 11 Types of Observational Data Through Geometric Conditioning and Mixed Resolutions

The NASA-IBM Lunar Foundation Model spans scales from 1 to 100 meters per pixel using nearly 2 million aligned lunar data samples. The model, dataset, and fine-tuning code are all publicly available, but its ice-related outputs remain indirect inferences and must not be used for landing or exploration decisions.

2026-09-11

LLMOps 與向量資料庫

Dify 1.17.1 Narrows Knowledge Base Key Permissions, but Bundled Weaviate Must Be Upgraded Sequentially Across 12 Versions

The latest Dify release adds dataset-scoped API keys and explicit database sessions, while shortening agent event retention to control Redis growth. Existing deployments using the bundled Weaviate cannot simply update the image; otherwise, objects may remain readable even as vector retrieval becomes permanently impaired.

2026-09-11

應用研究

TIER IV Open-Sources METEOR: One Network Uses Eight Cameras for Simultaneous Perception and Path Planning on Jetson Orin

The 54M-parameter METEOR builds a BEV representation from eight camera feeds and outputs 12 driving tasks in a single forward pass. Its INT8 version runs in approximately 67–70 ms on Jetson AGX Orin. The code, ONNX model, weights, and demo data are public, but the deployment version disables temporal memory, and no public autonomous-driving benchmark is yet available for cross-model comparison.

2026-09-10