ML Resources — July 13
Github Copilot is astonishingly good in many instances, and hilariously bad at several others. Open AI, the creators of Codex, the language model that powers Github Copilot, have released an evaluation framework for evaluating models on their ability to synthesize code from docstrings. Their framework includes a dataset of hand-written programming problems called ‘HumanEval’.
The best paper awardees for ACL 2021 have been announced. The ‘Best Paper’ award goes to Xu et al. for their paper on constructing a optimal vocabulary for learning that takes into account both vocabulary size and entropy. (Corpus entropy decreases as vocabulary size increases, which helps in model learning, while on the other hand, too many tokens causes token sparsity and inhibits model learning)
The ‘Best Theme Paper’ award is being awarded to Yin et al. , which is a position paper that discusses the linguistic properties of Signed Languages and points out current challenges in Sign Language Processing.
Six other papers are being awarded as ‘Outstanding Papers’, including Sinha et al.’s paper on their evidence that even state-of-the-art NLI models are quite invariant to random word-order permutations.
The University of Cambridge is currently organizing a free virtual summer school on the History of AI, with a list of speakers from around the world.
Aggregate Intellect
Aggregate Intellect is a Global Marketplace where ML Developers Connect, Collaborate, and Build. Connect with peers & experts at https://ai.science or Join our Slack Community.
- Check out the user generated Recipes that provide step by step, and bite sized guides on how to do various tasks
- Join our ML Product Challenges to build AI-based products for a chance to win cash prizes
- Connect with peers & experts through the ML Discussion Groups or Expert Office Hours
