I've been waiting for this 🤯
Open Source LLM Models Surpass GPT-3.5 🎉
In a groundbreaking development, a remarkable set of open-source LLM models has outperformed the capabilities of GPT-3.5.
What truly amazed me is not only the exceptional performance of these models but also the astonishing dataset they were trained on. Surprisingly, this dataset consisted of just 6,000 GPT4 conversations. 🤯
It's clear that high-quality data is working wonders. ✨
A few more details:
- OpenChat: Built upon LLAMA-13B, with a context length of 2048.
- Achieves an impressive 105.7% of the ChatGPT score on the Vicuna GPT-4 evaluation. 🏆
- Achieves a remarkable 80.9% win-rate on AlpacaEval. 🥇
- OpenChat-8192: Based on LLaMA-13B, with an extended context length of 8192.
- Achieves an outstanding 106.6% of the ChatGPT score on the Vicuna GPT-4 evaluation. 🌟
- Achieves an impressive 79.5% win-rate on AlpacaEval. 🥇
📚Links:
OpenChat on HF: https://t.co/ByuXE5zAwg
OpenChat 8192 on HF: https://t.co/6yRYpwsjYk
OpenCoderPlus: https://t.co/hiwF2gwoPC
Dataset: https://t.co/D43Kmn8seB
Code: https://t.co/REmURzPxku
Just out, the June 2023 IFAC NL issue. We are happy to share the ninth cartoon of this series. It provides some inspiration for (or insights on) state-of-the-art supervision practices. ;-) Thanks to @brianbdouglas https://t.co/5BoQkVuSKq
🤯🤯🤯Blown away by how fast @CollimatorAI has integrated generative AI using GPT-4 into their modeling framework.
Now possible to design and analyze control systems symbolically from text prompts!
Check out Collimator and control smarter! 🧠🧠🧠
https://t.co/mx67bHtl66
🎉Book PDF (730 pages) :
"High-Dimensional Data Analysis with Low-Dimensional Models" by Wright & Ma
👉PDF: https://t.co/IrhpIXZLSV
Preface by E. Candes
🎉🎉🎉
"Mathematics for machine learning" (CUP, 2020).
Free PDF from https://t.co/mbzGgyoAVP 😊
A must-read masterpiece, nicely structured and richly color-illustrated!
5 Books That Will Protect You From Being Manipulated
1. Games People Play
Explores psychological games where humans unconsciously manipulate each other to act alienatingly and self-destructively.
Browse over 500+ research cards online (or download PDF) covering machine learning, statistics, information geometry, and information theory among others!
Many tidbits and machine learning tricks...
👉https://t.co/7DkExOZZPi
I did a 12 mile run along the river in Boston today, while listening to Beginning of Infinity by David Deutsch, which is one of my favorite books. It's endlessly optimistic and inspiring. I'm grateful to be alive and to be on this idea exploration journey with all of you ❤️
Open-source LLMs released a week ago are already outdated. Not only are traditional leaderboard-style “benchmarks” being improved on, but new capabilities and efficiencies are also being unlocked at rapid pace.