Jacob Morrison
@jacobcaresPYI @ @allen_ai @ai2_allennlp
Similar User
@zhaofeng_wu
@gu_yuling
@hamishivi
@machelreid
@danieljwkim
@CMCarmenRivera
@def_jackie
@IanMagnusson
@mechanicaldirk
@PeterWestTM
@viviansong
@Neeeeens_
@TylerAdler
@jrfisher552
@officialhotdogs
I'm so excited that we're finally releasing Tülu 3, our new post-training recipe! We're releasing models built on top of Llama 3.1 base (OLMo coming soon!), all of our datasets, a (73 page!) paper, new evaluations, and all of our code.
TÜLU 3 Pushing Frontiers in Open Language Model Post-Training
Open Post-Training recipes! Some of my personal highlights: 💡 We significantly scaled up our preference data! (using more than 330k preference pairs for our 70b model!) 💡 We used RL with Verifiable Rewards to improve targeted skills like math and precise instruction following…
Long nights at the office were only possible thru the strength given by the Tülus around us. Starter pack: - brat tulu, a @jacobcares favorite - PNW tulu, don’t forget where @allen_ai is from - dank tulu 💪 - tulu at tulu, bc @faeze_brh told us tulu is farsi is sunrise
Excited to release Tulu 3! We worked hard to try and make the best open post-training recipe we could, and the results are good! I was lucky enough to work on almost every stage of the pipeline in one way or another. Some comments + highlights ⬇️
Meet Tülu 3 -- a set of state-of-the-art instruct models with fully open data, eval code, and training algorithms. We invented new methods for fine-tuning language models with RL and built upon best practices in the community to scale synthetic instruction and preference data.…
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 📜arxiv.org/abs/2409.02060 🧵1/9
This project was a lot of fun! I learned a lot about reward models in the process, and we hope the community can build on RewardBench to learn even more. Congrats to the Starling team too 🌟🐦⬛ Let us know if there are any other interesting domains you think we should cover!
Excited to share something that we've needed since the early open RLHF days: RewardBench, the first benchmark for reward models. 1. We evaluated 30+ of the currently available RMs (w/ DPO too). 2. We created new datasets covering chat, safety, code, math, etc. We learned a lot.…
There are only 10 days left to apply to the Predoctoral Young Investigator program with AllenNLP! Don't miss your chance to work with world-class researchers creating AI for the common good. Applications must be submitted by January 15th --> boards.greenhouse.io/thealleninstit…
hey @emnlpmeeting #EMNLP2023 i lost my bucket hat 😢 it’s white w yellow bananas 🍌 on it like in my profile, pls DM me if found 🙏🏻
Thanks for the shoutout Yu! I'm hiring multiple PhD students this cycle. If you are interested in building NLP solutions for real use cases that work for all people, those that can uniformly support diverse languages, domains, populations, and individuals, please apply.
Are you looking to be part of AI for the common good while preparing for a PhD program? Consider a 1-3 year-long stint as a Predoctoral Young Investigator! Apply by Jan. 15th 2024 --> boards.greenhouse.io/thealleninstit…
We're taking applications for Predoctoral Young Investigators! Prepare for a PhD program and partner with top mentors on cutting edge NLP research. Apply by Jan. 15th, 2024 --> boards.greenhouse.io/thealleninstit…
Please forward this flyer if you are in the Austin area. I really appreciate all of your help.
Since our #Cinerama acquisition, many of you have asked how you can help 🎦 A gift to our annual fund is a great way. Think of it as an investment in choc popcorn! You're also solidifying our future & our efforts to preserve the cinema experience. Give at siff.info/supportsiff-tw.
United States Trends
- 1. Thanksgiving 2,4 Mn posts
- 2. Bears 61,9 B posts
- 3. Caleb Williams 5.888 posts
- 4. Thankful 462 B posts
- 5. #IDEGEN N/A
- 6. Gibbs 13,4 B posts
- 7. Eberflus 6.185 posts
- 8. Turkey 300 B posts
- 9. #CHIvsDET 5.283 posts
- 10. Ben Johnson 2.754 posts
- 11. #OnePride 8.858 posts
- 12. Penei Sewell 2.462 posts
- 13. Goff 6.866 posts
- 14. #NationalDogShow N/A
- 15. DJ Moore 2.068 posts
- 16. #Picksgiving 6.644 posts
- 17. Tyrique Stevenson N/A
- 18. Laporta 5.306 posts
- 19. Tony Romo N/A
- 20. Grateful 284 B posts
Who to follow
-
Zhaofeng Wu
@zhaofeng_wu -
Yuling Gu (applying to PhD)
@gu_yuling -
Hamish Ivison
@hamishivi -
Machel Reid
@machelreid -
Joongwon Kim
@danieljwkim -
CM Carmen Rivera
@CMCarmenRivera -
Jackie Chan
@def_jackie -
Ian Magnusson
@IanMagnusson -
Mechanical Dirk
@mechanicaldirk -
Peter West
@PeterWestTM -
Vivian Song 宋瑋雯
@viviansong -
Nina Kohn🌺
@Neeeeens_ -
Lur
@TylerAdler -
Jillian Fisher
@jrfisher552 -
Sparkling Hot Dog Water
@officialhotdogs
Something went wrong.
Something went wrong.