TGTGInsighttelegram intelligenceLIVE / telegram public index
← Linuxgram 🐧

TGINSIGHT SIMILAR POSTS

Sib kontenut simili

Kanal tas-sors @linuxgram · Post #17841 · Fra 19

📰 AI Helped Uncover A "50-80x Improvement" For Linux's IO_uring Linux block maintainer and IO_uring lead developer Jens Axboe recently was debugging some slowdowns in the AHCI/SCSI code with IO_uring usage. When turning to Claude AI to help in sorting through the issue, patches were devised that can deliver up to a "literally yield a 50-80x improvement on the io_uring side for idle systems." The code is on its way to the Linux kernel... 🔗 Source: https://www.phoronix.com/news/AI-50-80x-IO-uring #linux#kernel

Riżultati

1 post simili nstab

Tfittxija: #openweights

当前筛选 #openweights清除筛选
Machinelearning

@ai_machinelearning_big_data · Post #8680 · 02/10/2025 17:01

✔️IBM представила Granite 4.0 — новое семейство open-weights языковых моделей от 3B до 32B параметров. Четыре новые модели: - Granite 4.0 H Small - 32B/9B активных параметров - Granite 4.0 H Tiny - 7B/1B - Granite 4.0 H Micro - 3B/3B - Granite 4.0 Micro - 3B/3B Benchmarking (Artificial Analysis Index): - Granite 4.0 H Small: 23 балла (на 8 выше Granite 3.3 8B), обходит Gemma 3 27B (22), но уступает Mistral Small 3.2 (29) и Qwen3 30B A3B (37). - Granite 4.0 Micro: 16 баллов, выше Gemma 3 4B (15) и LFM 2 2.6B (12). ⚡ Token efficiency: - Granite 4.0 Small — 5.2M токенов - Granite 4.0 Micro — 6.7M токенов Обе модели заметно эффективнее Granite 3.3 8B и большинства non-reasoning моделей <40B. Детали: - Контекст: до 128K токенов - Лицензия: Apache 2.0 - Granite 4.0 H Small доступна на Replicate по $0.06 / $0.25 за 1M input/output токенов - Все модели доступны на Hugging Face - Модель Micro (3.4B) можно запускать полностью локально. 🔗 Hugging Face: https://huggingface.co/collections/unsloth/granite-40-68ddf64b4a8717dc22a9322d 🔗Unsloth: https://docs.unsloth.ai/new/ibm-granite-4.0 @ai_machinelearning_big_data #AI#IBM#Granite4#LLM#OpenWeights