Luca Soldaini 🎀
soldaini.net
I like tokens! Leading the Olmo data team at @ai2.bsky.social with @kylelo.bsky.social // all opinions sampled from my own stochastic parrot more at https://soldaini.net
👤 Joined on 04 March 2023
8815
Followers
485
Following
410
Total Posts
📜 Kullanıcı Geçmişi
- 09 Mar 2023, 23:18
🎉 Bluesky'e katıldı - İlk kullanıcı adı:
@soldni.bsky.social - 30 Nis 2023, 17:24
✏️ Kullanıcı adı değişti:
@soldni.bsky.social→@soldaini.net - 02 Tem 2023, 17:45
✏️ Kullanıcı adı değişti:
@soldaini.net→@soldni.bsky.social - 05 Eki 2023, 17:44
✏️ Kullanıcı adı değişti:
@soldni.bsky.social→@soldaini.net - 10 Kas 2023, 04:15
🔄 Profil bilgileri güncellendi
🚫 Blokladığı Kullanıcılar
❤️ Beğendiği Gönderiler
We’re updating olmOCR, our model for turning PDFs & scans into clean text with support for tables, equations, handwriting, & more. olmOCR 2 uses synthetic data + unit tests as verifiable rewards to reach state-of-the-art performance on challenging documents. 🧵
Show Post
How much data people use to train tokenizers nowadays? Trying to figure out, but so often people just use a trained tokenizer, so a bit tricky cc @soldaini.net and the OLMo folks
Show Post
Giving a talk at NYU text as data tomorrow (Thursday) at 11! Catch me there buds
Show Post