🚀A high-speed, fully local voice input tool exclusively for Mac, optimized for Apple Silicon (Metal)
localで完結した音声入力アプリを作りました。MLX-Whisperに対応でApple Silicon搭載のMacで20秒程度の読み上げを2秒前後で行えます。
https://t.co/5vYIQW0WQn
Papers like these are important for people competing in big reasoning competitions like AIMO or ARC-AGI.
The problem is that if one takes a closer look, there are some issues with the impressive claims:
- MATH is an outdated benchmark by now
- the numbers don't add up. The last sentence on page 1 states "Qwen-2.5-7B-Instruct improves from 76% to 95% while training just 10,000 parameters". This conflicts with table 2, which in turn is also unclear, as the parameter count doesn't seem to match with the # column.
NVIDIA just released Surgical Qwen2.5-VL on Hugging Face!
This new multimodal LLM is fine-tuned to recognize surgical actions, instruments, and targets directly from endoscopic video frames.
A huge step for surgical workflow analysis.
https://t.co/XEVJuxUyD2
KaggleをMulti-agentで解くぞ論文。Playgroundとかの簡単なテーブルデータしか使ってないけど、ツヨツヨKaggler集めればガチコンペでも(部分的に)けっこう役立つAgent作れる気はする👀
AutoKaggle: A Multi-Agent Framework for Autonomous Data Science Competitions
https://t.co/vmNGvziaFF