Gabriele Sarti

232 Followers
343 Following
27 Posts
PhD Student in Explainable NMT at @gronlp & Core Dev @inseq. Interpretability ∩ #NLG #NLProc. Prev: AWS AI, Aindo, ItaliaNLP Lab
LocationGroningen, Netherlands
Websitehttps://gsarti.com
Twitterhttps://twitter.com/gsarti_
Githubhttps://github.com/gsarti

After a year of restless development, I'm finally happy to announce Inseq, a new tool to democratize post-hoc #interpretability of sequence #generation models 🐛 https://github.com/inseq-team/inseq #nlproc #xai

Inseq supports most Hugging Face decoder-only and seq2seq models, with a variety of attribution methods already baked in, and many more to come! 🚀 Visualizing MetaAI's Galactica writing LaTeX or GoogleAI Flan-T5 doing commonsense reasoning takes only 3 lines of code 🤯

GitHub - inseq-team/inseq: Interpretability for sequence generation models 🐛 🔍

Interpretability for sequence generation models 🐛 🔍 - GitHub - inseq-team/inseq: Interpretability for sequence generation models 🐛 🔍

GitHub
Not quite the most plausible explanation, but still mind-blowingly fluent 🤯 hats off to OpenAI