iiinova — where your digital self keeps evolving

Luca Soldaini 🎀 @soldaini.net · 22/10/2025

it’s OCR week! learn how we use verifiable rewards against unit tests to improve olmOCR’s PDF understanding state of the art OCR, fully open model:

1232

Luca Soldaini 🎀 @soldaini.net · 12/09/2025

which one of you im gonna have the pleasure to see at COLM???

5130

Luca Soldaini 🎀 @soldaini.net · 20/08/2025

my keystrokes go though light-up starry cable OF COURSE my code is better than yours

0150

Luca Soldaini 🎀 @soldaini.net · 05/08/2025

12+ years in this country, first time I get to wear this sticker 🗳️

0160

Luca Soldaini 🎀 @soldaini.net · 18/07/2025

wearing italian camo* at ICML *ordering an ice lattes rather than espressos at coffee shops

140

Luca Soldaini 🎀 @soldaini.net · 23/06/2025

new @ai2.bsky.social office has something for everyone: stunning views for the outdoorsy kind, 2.5 Gbps connection at every desk for the indoor nerds

0130

Luca Soldaini 🎀 @soldaini.net · 18/06/2025

Waymo is cool but BART from SFO to downtown SF is cooler 101 can be as dark red as you want on google maps!

080

Luca Soldaini 🎀 @soldaini.net · 06/06/2025

2025 AI hot take: everyone should use FastText more. Word embeddings are awesome.

3253

Luca Soldaini 🎀 @soldaini.net · 01/06/2025

today might be rainy, but PNW summer is already here

0120

Luca Soldaini 🎀 @soldaini.net · 15/05/2025

I've silenced all notifications on all my devices and it's truly the best thing ever ...I am considering allowing calendar notifications tho cuz I almost missed 3 meetings already 😅

2240

Luca Soldaini 🎀 @soldaini.net · 27/04/2025

two weeks traveling and I miss my mechanical keyboard so much

140

Luca Soldaini 🎀 @soldaini.net · 22/04/2025

when someone says they wanna bring me to their favorite italian restaurant

0150

Luca Soldaini 🎀 @soldaini.net · 02/04/2025

I am still perpetually in awe that skill emergence exists in language models million of caveats but we have models that pick up capabilities from plain text??? it's so magical, I can't believe we got such treat

0222

Luca Soldaini 🎀 @soldaini.net · 26/03/2025

bluesky deserves to know we’ve adopted a cat and he’s the most handsome boy

2620

Luca Soldaini 🎀 @soldaini.net · 20/03/2025

Summary of our recommendation we submitted to White House to ensure success of open & transparent AI As a meta point, I’m very grateful to be in a position where I can put my technical expertise in the service of policy needs 🥰

070

Luca Soldaini 🎀 @soldaini.net · 10/03/2025

"excuse me sir do you have a moment to talk about olmOCR"

2292

Luca Soldaini 🎀 @soldaini.net · 06/03/2025

if i ever find a genie lamp im gonna save a wish for “uninvent pydantic and click” 😤

360

Luca Soldaini 🎀 @soldaini.net · 20/02/2025

in the upcoming LLMs war, i choose a neutral team

1360

Luca Soldaini 🎀 @soldaini.net · 18/02/2025

This was such a fun project to work on! We release efficient classifiers 🌐 to partition large corpora, and use them to improve sampling for LLM pretraining great work lead by @awettig.bsky.social 👇

0182

Luca Soldaini 🎀 @soldaini.net · 16/02/2025

my toxic trait is enjoying the AWS cost optimization dashboard a little too much

1210

Reposted by Luca Soldaini 🎀

Kyle Lo @ ICML2026 🇰🇷 @kylelo.bsky.social · 11/02/2025

one benefit of open models is privacy 🔐you can run them locally & keep all your data on your local device instead of sending it to a company elsewhere congrats @soldaini.net for heavy lifting, showing our OLMoE model can run on iPhones📱

0173

Reposted by Luca Soldaini 🎀

Nathan Lambert @natolambert.bsky.social · 11/02/2025

Best part of this that Luca isn’t highlighting to start is that we trained a way better OLMoE for this too. All from better annealing and post train. Didn’t need to redo pre training. Goes to show how much potential these models have! new instruct model: huggingface.co/allenai/OLMo...

1174

Reposted by Luca Soldaini 🎀

Costa Huang @vwxyzjn.bsky.social · 11/02/2025

🤯 Check out our new iOS OLMoE app that runs the model on-device! We also trained new OLMoE-1B-7B-0125 this time using the Tulu 3 recipe. Very exciting that RLVR improved gsm8k by almost 10 points for OLMoE 🔥 A quick 🧵

151

Luca Soldaini 🎀 @soldaini.net · 11/02/2025

They made me do video 😬 but for a good reason! We are launching an iOS app–it runs OLMoE locally 📱 We're gonna see more on-device AI in 2025, and wanted to offer a simple way to prototype with it App: apps.apple.com/us/app/ai2-o... Code: github.com/allenai/OLMo... Blog: allenai.org/blog/olmoe-app

54913

Luca Soldaini 🎀 @soldaini.net · 10/02/2025

dreading "how was your weekend" questions don't want to tell people I spent my sunday recompiling vLLM from scratch

1110

Luca Soldaini 🎀 @soldaini.net · 10/02/2025

three things are certain in life: death, taxes, and Claude switching to concise mode during US business hours

2372

Luca Soldaini 🎀 @soldaini.net · 07/02/2025

ok but how many LLMs you know are available on BluRay

3414

Reposted by Luca Soldaini 🎀

Ai2 @ai2.bsky.social · 30/01/2025

Here is Tülu 3 405B 🐫 our open-source post-training model that surpasses the performance of DeepSeek-V3! It demonstrates that our recipe, which includes RVLR scales to 405B - with performance on par with GPT-4o, & surpassing prior open-weight post-trained models of the same size including Llama 3.1.

29221

Luca Soldaini 🎀 @soldaini.net · 30/01/2025

Do people have a sandboxed code execution setup for LLM verifiers they like? My current prototype is Docker + Google Cloud Run; if I keep cache warm, startup is about 600 ms. Firewall to block any network egress from container. Would appreciate something with lower latency

230

Luca Soldaini 🎀 @soldaini.net · 28/01/2025

calling DeepSeek R1 via APIs (tried Together AI, DeepInfra, DeepSeek APIs, our own vLLM server) is a somber reminder of how painful thinking-MoEs are at scale at best ~20TPS, while trying to generate 16k-32k per request really shows how impressive/hard O1 or O3 deployment is

191

Luca Soldaini 🎀 @soldaini.net · 27/01/2025

it's gonna be a sad day when thinking models will figure out ASCII art, current capabilities crack me up so much:

4221

Luca Soldaini 🎀 @soldaini.net · 20/01/2025

can someone familiar with Chinese AI space map Moonshot AI/Baichuan/MiniMax/Ziphu AI vs DeepSeek/Qwen for me? What about InternLM? I know models/papers but I have no sense of relative scale of companies 🙏

040

Luca Soldaini 🎀 @soldaini.net · 13/01/2025

just learned how gorgeous the old Seattle Times building was, so sad it got demolished for another downtown tower

040

Reposted by Luca Soldaini 🎀

Kyle Lo @ ICML2026 🇰🇷 @kylelo.bsky.social · 03/01/2025

kicking off 2025 with our OLMo 2 tech report while payin homage to the sequelest of sequels 🫡 🚗 2 OLMo 2 Furious 🔥 is everythin we learned since OLMo 1, with deep dives into: 🚖 stable pretrain recipe 🚔 lr anneal 🤝 data curricula 🤝 soups 🚘 tulu post-train recipe 🚜 compute infra setup 👇🧵

26917

Reposted by Luca Soldaini 🎀

Nathan Lambert @natolambert.bsky.social · 03/01/2025

Everyone wants open-source language models but no one wants to lift these heavy ass weights. We just released our paper "2 OLMo 2 Furious" Can't stop us in 2025. Links below.

65610

Luca Soldaini 🎀 @soldaini.net · 03/01/2025

OLMo 2 tech report is out! We get in the weeds with this one, with 50+ pages on 4 crucial components of LLM development pipeline:

3479

Luca Soldaini 🎀 @soldaini.net · 30/12/2024

looking for new color themes for my terminal and it's hard to pass on this one

2160

Luca Soldaini 🎀 @soldaini.net · 26/12/2024

rough seas but it was all so worth it 🌊🚢🥰

1190

Luca Soldaini 🎀 @soldaini.net · 20/12/2024

Anthropic/GDM could do the funniest thing and nam their next thing Chlorofluorocarbon

1101

Luca Soldaini 🎀 @soldaini.net · 20/12/2024

Anthropic, OpenAI: mid-century revival Meta, Google: post Corporate Memphis, big-tech minimalism I'm ready to give my money to the first brutalist AI endeavor 😍

1180

Luca Soldaini 🎀 @soldaini.net · 19/12/2024

"AI is great, but the sequel AI2 is even better" - @hugochoss.bsky.social holiday vibes hitting hard at the office rn 🎅🎄

2100

Luca Soldaini 🎀 @soldaini.net · 18/12/2024

1-800 number for chatgpt is a fun toy/marketing just enjoy it for what it is no reason to be grumpy around the holidays

070

Luca Soldaini 🎀 @soldaini.net · 18/12/2024

where can i download LMSYS / ChatArena historical data? looking for ELO/ranks week by week. Someone made this interactive viz a while back, would like to recreate: public.flourish.studio/visualisatio...

public.flourish.studio

Top LLMs - Chatbot arena - May - Dec 2023

A Flourish data visualization by Peter Gostev

180

Luca Soldaini 🎀 @soldaini.net · 13/12/2024

new achievement: ✅ cited a bluesky post in a paper

3523

Luca Soldaini 🎀 @soldaini.net · 13/12/2024

fella keeps staring at me no matter how much i tell him in not interested in their AI startup

1120

Luca Soldaini 🎀 @soldaini.net · 11/12/2024

google slides refreshed their templates and i'm lowkey impressed with how unhinged they are

0412