srush_nlp's profile picture. Researcher at Cursor
https://www.youtube.com/@srush_nlp

Sasha Rush

@srush_nlp

Researcher at Cursor https://www.youtube.com/@srush_nlp

Sasha Rush reposted

Was super fun to organize this workshop!! Thanks everyone: speakers, panelists, audience. facebookresearch.github.io/RAM/workshop/C…

jaseweston's tweet image. Was super fun to organize this workshop!! Thanks everyone: speakers, panelists, audience.
facebookresearch.github.io/RAM/workshop/C…

Sasha Rush reposted

Here's the slides from my talk on "Three Lessons from DeepSeek-R1" that I gave at #COLM2025 docs.google.com/presentation/d… It was fun to review how our collective understanding of what works / doesn't work for reasoning models has evolved in a short few months!

When you unintentionally mirror your memes

_lewtun's tweet image. When you unintentionally mirror your memes


Sasha Rush reposted

Heading to Montreal for the #COLM2025 afterparty


Sasha Rush reposted

Talk from Wenting Zhao of Qwen on their plans during COLM. Seems like 1 word is the plan still: scaling training up! Let’s go.

natolambert's tweet image. Talk from Wenting Zhao of Qwen on their plans during COLM. Seems like 1 word is the plan still: scaling training up! Let’s go.

🤗

Highlight of the day: met the @huggingface team, got a selfie and a signed book from @lvwerra. @_lewtun 🙌🫶

Amirhosein_Gh0's tweet image. Highlight of the day: met the @huggingface team, got a selfie and a signed book from @lvwerra. @_lewtun 🙌🫶


Sasha Rush reposted

Happening today! If you are at #COLM2025, come by the Workshop on the Application of LLM Explainability to Reasoning and Planning at 2:40 ET to see my talk on challenges in human-agent communication and how the interpretability community can help address them!

📢Schedule is now finalized! Join the brilliant invited talks with us (…reasoning-planning-workshop.github.io): Greg Durrett @gregd_nlp : LLM Reasoning Beyond Scaling Huan Sun @hhsun1 : How Explanations Can Advance Capability and Safety: World Modeling and Circuit Discovery Ana Marasović…



Sasha Rush reposted

Today at COLM, Cohere Labs Sr Research Scientist, Julia Kreutzer will be presenting at 2 workshops. First, the Multilingual Data Quality Signals workshop, bringing together researchers across disciplines to discuss & present research on data quality signals in multilingual data.

Cohere_Labs's tweet image. Today at COLM, Cohere Labs Sr Research Scientist, Julia Kreutzer will be presenting at 2 workshops.

First, the Multilingual Data Quality Signals workshop, bringing together researchers across disciplines to discuss & present research on data quality signals in multilingual data.

Sasha Rush reposted

Proud advisor moment at #COLM2025! Congrats to all the organizers for a wonderful week. I’m ready for COLM 3…but first workshops and then back to the West Coast Monday where I’ll be speaking at Tech St Santa Monica for LA Tech Week.

GabrielSaadia's tweet image. Proud advisor moment at #COLM2025! Congrats to all the organizers for a wonderful week. I’m ready for COLM 3…but first workshops and then back to the West Coast Monday where I’ll be speaking at Tech St Santa Monica for LA Tech Week.

Sasha Rush reposted

COLM is getting schmidhhubered by none other than @SchmidhuberAI about 10 years of progress #COLM2025 #RAM2 workshop Probably the only presentation at COLM with no slides.

sivareddyg's tweet image. COLM is getting schmidhhubered by none other than @SchmidhuberAI about 10 years of progress #COLM2025 #RAM2 workshop 

Probably the only presentation at COLM with no slides.

Sasha Rush reposted

Excited about this! We’re getting over 500 TPS on Blackwell with DeepSeek-V3.1. The more you use it, the greater the speedup. Blog post: together-ai.webflow.io/blog/adaptive-…

This work, led by @_junxiong_wang and @ben_athi, is a first step towards building AI systems that evolve and get better as you use them. More to come!



Sasha Rush reposted

Excited to give a talk at the interplay workshop tomorrow! Come say hi! Alas, it’s my only day at COLM. Catch me at the coffee breaks or the roundtable.

✨ The schedule for our INTERPLAY workshop at COLM is live! ✨ 🗓️ October 10th, Room 518C 🔹 Invited talks from @sarahwiegreffe @johnhewtt @amuuueller @kmahowald 🔹 Paper presentations and posters 🔹 Closing roundtable discussion. Join us in Montréal! @COLM_conf

interplaywrkshp's tweet image. ✨ The schedule for our INTERPLAY workshop at COLM is live! ✨
🗓️ October 10th, Room 518C
🔹 Invited talks from @sarahwiegreffe @johnhewtt @amuuueller @kmahowald 
🔹 Paper presentations and posters 
🔹 Closing roundtable discussion.

Join us in Montréal! @COLM_conf


Sasha Rush reposted

happening now! 🐏

kchonyc's tweet image. happening now! 🐏

will be at the RAM workshop 2.0 today, partying like 2015! 🥳 #COLM2025

kchonyc's tweet image. will be at the RAM workshop 2.0 today, partying like 2015! 🥳

#COLM2025


Sasha Rush reposted

Two power packed panels on AI Safety and hundreds of audience at @ServiceNowRSRCH, Montreal. #COLM2025 Topics covered: Open source and safety Safety as a system vs standalone Deceptiveness Chain of thought faithfulness Sandbagging Emergent misalignment with fine-tuning User…

sivareddyg's tweet image. Two power packed panels on AI Safety and hundreds of audience at @ServiceNowRSRCH, Montreal. #COLM2025

Topics covered:
Open source and safety 
Safety as a system vs standalone
Deceptiveness
Chain of thought faithfulness
Sandbagging
Emergent misalignment with fine-tuning
User…
sivareddyg's tweet image. Two power packed panels on AI Safety and hundreds of audience at @ServiceNowRSRCH, Montreal. #COLM2025

Topics covered:
Open source and safety 
Safety as a system vs standalone
Deceptiveness
Chain of thought faithfulness
Sandbagging
Emergent misalignment with fine-tuning
User…

Sasha Rush reposted

Fluid LM benchmarking from @vjhofmann and @allen_ai #COLM2025 I've already shilled this paper, it's great. With item level difficulty (IRT model) you can estimate *latent capability* of a model rather than raw performance by giving it samples that maximize information gain

m2saxon's tweet image. Fluid LM benchmarking from @vjhofmann and @allen_ai #COLM2025 

I've already shilled this paper, it's great. With item level difficulty (IRT model) you can estimate *latent capability* of a model rather than raw performance by giving it samples that maximize information gain

Sasha Rush reposted

🧵Come check out our Thursday morning poster at @COLM_conf - 11AM - Poster #15! “Mitigating Modal Imbalance in Multimodal Reasoning” (1/n)

neilkale's tweet image. 🧵Come check out our Thursday morning poster at @COLM_conf - 11AM - Poster #15! 

“Mitigating Modal Imbalance in Multimodal Reasoning” (1/n)

Nathan is going to give this talk today @ 2pm at COLM in room 524C. Should be really interesting.

I gave a talk today at The Curve on the state of open models. Here are the slides, recording soon. Topics include: Chinese ecosystem, reflections on DeepSeek, the demise of Llama, who will fill the U.S. market, what local models do, ATOM project & ai2, and more topics

natolambert's tweet image. I gave a talk today at The Curve on the state of open models.
Here are the slides, recording soon.

Topics include: Chinese ecosystem, reflections on DeepSeek, the demise of Llama, who will fill the U.S. market, what local models do, ATOM project & ai2, and more topics


Sasha Rush reposted

If you are at COLM 2025 and interested in code gen, agents, reasoning and or infra, Reach out to join an @a16z x @cursor_ai luncheon tomorrow (Friday) near the conference Cc @chsrbrts @srush_nlp @StringChaos


Sasha Rush reposted

made it to #COLM2025

kchonyc's tweet image. made it to #COLM2025

Loading...

Something went wrong.


Something went wrong.