Ai2
@allen_ai
Followers
59,077
Following
370
Media
369
Statuses
2,588
Breakthrough AI to solve the world's biggest problems. › Join us: › Newsletter:
Seattle, WA
Joined September 2015
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
Bivol
• 126829 Tweets
Oregon
• 90347 Tweets
#AEWWrestleDream
• 81370 Tweets
Hiro
• 60801 Tweets
emilia
• 52629 Tweets
Ohio State
• 51405 Tweets
Howard
• 44183 Tweets
Dennis Quaid
• 41427 Tweets
Tennessee
• 40577 Tweets
ワコール
• 31418 Tweets
Valencia
• 30120 Tweets
GALA EN PAPIS
• 28305 Tweets
MÁS CERCA DE KARIME
• 21477 Tweets
Ole Miss
• 19531 Tweets
イッテQ
• 15531 Tweets
Ryan Day
• 13540 Tweets
セブンのレジ
• 13471 Tweets
Buckeyes
• 12442 Tweets
フリーナ
• 12335 Tweets
マッドレックス様
• 11560 Tweets
さつまいもの日
• 10712 Tweets
Napier
• 10408 Tweets
Pinned Tweet
Meet Molmo: a family of open, state-of-the-art multimodal AI models.
Our best model outperforms proprietary systems, using 1000x less data.
Molmo doesn't just understand multimodal data—it acts on it, enabling rich interactions in both the physical and virtual worlds.
Try it
50
272
1K
Ai2 Retweeted
New work from
@allen_ai
and
@UW
: SOTA multimodal AI models that are truly open-source.
The authors have open-sourced their full pipeline, including model weights (Molmo), datasets (PixMo), and training code.
Excited to have
@inkynumbers
and team discussing their work here!
2
7
25
Excited for our CEO Ali Farhadi to be back for
@MadronaVentures
' 3rd annual
#IASummit
tomorrow in Seattle, this time on a fantastic panel with
@aidangomez
, Asha Sharma, and
@EricNewcomer
– looking forward to it!
1
3
10
Ai2 Retweeted
Introducing Molmo: a family of open state-of-the-art multimodal language models that push the frontier of vision by pointing to reason about images!
The data curation has been excellent and is leading to our ridiculously tiny 7B models to outperform on par or surpass top models
14
42
256
Ai2 Retweeted
Delighted to share Molmo!
With an extreme focus on data quality over quantity, Molmo is fast to train, competitive with every model out there (benchmarks and Elo!) and fully open source!
Open matches closed.
Less can be better than more.
Small rivals big!
8
21
126
Ai2 Retweeted
Open-source models beating closed models will become more and more common. Scaling has diminishing returns. The best solution will not have the largest scale but best approach or data. Especially with test-time compute, you do not need the best model to have the best solution.
12
80
511
We're always excited to celebrate
#OpenScienceWeek
with
@AllenInstitute
! From our family of models to the data that trains them, we're opening up AI so we can all understand this new tech better. Revisit our
@axios
talk w/
@inafried
on this topic:
1
2
3
Our friends at
@SCSatCMU
have just launched a new podcast, Does Compute, which features our visiting research scientist
@MaartenSap
in episode two, "What AI Isn't, Part 2" —
3
4
17
Ai2 Retweeted
Next week on Wednesday I'll be talking about Open-Instruct and Tulu on the PyTorch youtube channel:
I'll talk about how open-instruct has changed to fit the needs of the Ai2 post-training team over the past year. Maybe some tulu 3 preview results too :)
0
17
56
Ai2 Retweeted
We’re proud to share our latest research, led by our own
@Muennighoff
and in partnership with
@allen_ai
: Introducing OLMoE, a best-in-class fully open source mixture-of-experts (MoE) language model with 1B active parameters that beats comparable LLMs and rivals many larger
1
10
45
Ai2 Retweeted
Releasing OLMoE - the first good Mixture-of-Experts LLM that's 100% open-source
- 1B active, 7B total params for 5T tokens
- Best small LLM & matches more costly ones like Gemma, Llama
- Open Model/Data/Code/Logs + lots of analysis & experiments
📜
🧵1/9
23
234
964
🐣Welcome the newest member to the OLMo family, OLMoE! This Mixture-of-Experts model is 100% open — it's efficient, performant, and ready for experimentation. Learn more on our blog:
2
16
78
AI reflects human biases — like covert racism. Our team's work highlights how LLMs output racist stereotypes about speakers of African American English (AAE), and is featured in
@Nature
today.
Beyond excited to share that this is now out in
@Nature
!
We show that despite efforts to remove overt racial bias, LLMs generate covertly racist decisions about people based on their dialect.
Joint work with amazing co-authors
@ria_kalluri
,
@jurafsky
, and Sharese King.
7
115
346
2
4
16
“Would I rather live in a world where there are many practitioners who can fix AI models when an attack or misuse happens, or where I’m at the mercy of a couple of institutes?” Our CEO Ali spoke w/
@Marketplace
about the importance of openness in AI:
1
5
16
We're so grateful to have passionate folks like Eli on our team who help us celebrate these big moments.🥳
0
0
5