kael-odin/2080ti-22G-kvmem-qwen3.8-27B-256K-36.6toks

RTX 2080 Ti 22G（sm_75）单卡本地长上下文最优解：KVMem 主线 262K 上下文 / 36.6 tok/s 实测（IQ3+MTP+视觉），KV 流式无损备档；完整源码复现链 + 构建/补丁/测试工具

GitHub repository with 5 stars and 1 forks.

Language: Python

Topics: cuda, inference, kv-cache, kvmem, llama-cpp, llm, local-llm, long-context, mtp, rtx-2080-ti

Open provider repository

24h trend summary

Trending score 0.08, freshness score 0.58, stars gained +0, forks gained +1.

Latest metric snapshot

2026-10-02: 5 stars and 1 forks.

Trending in Python

  1. 1. debpalash/VoiceStudio

    VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages.

    GitHub repository with 51,710 stars and 5,753 forks.

    Trending score: 6.75; stars gained: +3,514; forks gained: +325.

    Language: Python

    Topics: ai, audiobook, cuda, dubbing, elevenlabs-alternative, huggingface

  2. 2. vectorize-io/hindsight

    Hindsight: Agent Memory That Learns

    GitHub repository with 44,557 stars and 5,864 forks.

    Trending score: 6.49; stars gained: +1,517; forks gained: +93.

    Language: Python

    Topics: agentic-ai, agents, ai-memory, memory

  3. 3. NandhaKishorM/laya

    Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.

    GitHub repository with 30,099 stars and 2,624 forks.

    Trending score: 6.10; stars gained: +1,267; forks gained: +88.

    Language: Python

    Topics: calibration, classification, decision-model, huggingface, jev, modernbert

  4. 4. rohitg00/ai-engineering-from-scratch

    Learn it. Build it. Ship it for others.

    GitHub repository with 62,595 stars and 10,696 forks.

    Trending score: 6.00; stars gained: +1,156; forks gained: +156.

    Language: Python

    Topics: agents, ai, ai-agents, ai-engineering, computer-vision, course

  5. 5. tigerless-labs/autoharness

    Autoharness — a self-learning skill layer for Claude Code — distills skills from your real sessions, updates them as you work, and prunes the ones that stop getting used. No daemon, no benchmark.

    GitHub repository with 7,237 stars and 564 forks.

    Trending score: 5.73; stars gained: +831; forks gained: +84.

    Language: Python

    Topics: agent-skills, claude-code, claude-code-plugin, llm-agents, python

  6. 6. latent-spaces/brag

    You built it. Now brag. Turn the project you just created into a short, shareable launch video with one command.

    GitHub repository with 12,957 stars and 899 forks.

    Trending score: 5.65; stars gained: +747; forks gained: +61.

    Language: Python

    Topics: agent-skills, claude-code, claude-skills, hyperframes, launch-video, video-generation