iiinova — where your digital self keeps evolving

Neel Rajani

@neelrajani.bsky.social

640 followers 485 following 22 posts

PhD student in Responsible NLP at the University of Edinburgh, curious about interpretability and alignment

Posts Replies Media

Reposted by Neel Rajani

Rayo Verweij @rayo.dev · 17/06/2026

Absolute joy to present with Sarah Immel the paper we wrote with @neelrajani.bsky.social - Stepping Into the Black Box: Opening Up LLMs for Public Exploration Through Discursive Design at @acm-dis.bsky.social in Singapore. And we won an Honourable Mention!

Sarah, Rayo, and the session chair posing for the camera holding the Honourable Mention award. Above them, a projector screen says "Stepping into the Black Box: Opening Up LLMs to Public Exploration Through Discursive Design"

242

Reposted by Neel Rajani

Zeerak Talat زیرک طلعت (they/them) @zeerak.bsky.social · 12/06/2026

NLP reviews suck! But why is not clear– @aclrollingreview.bsky.social provides *a lot* of guidance for how to review, but in that, first principles get lost. So @adamlopez.bsky.social and I have written down some of our thoughts on first principles.

medium.com

The Missing First Principles of Reviewing for ACL

Zeerak Talat & Adam Lopez, University of Edinburgh

1236

Reposted by Neel Rajani

Aryo Pradipta Gema @aryopg.bsky.social · 02/05/2025

MMLU-Redux just touched down at #NAACL2025! 🎉 Wish I could be there for our "Are We Done with MMLU?" poster today (9:00-10:30am in Hall 3, Poster Session 7), but visa drama said nope 😅 If anyone's swinging by, give our research some love! Hit me up if you check it out! 👋

01711

Neel Rajani @neelrajani.bsky.social · 11/04/2025

Come say hi :)

010

Neel Rajani @neelrajani.bsky.social · 08/12/2024

Caught off-guard by the Llama 3.3 release? This is the loss of Llama-3.3-70B-Instruct (4bit quantized) on its own Twitter release thread. It really didn't like ' RL' (loss of 13.47) and wanted the text to instead go "... progress in online learning, which allows the model to adapt"

020

Neel Rajani @neelrajani.bsky.social · 03/12/2024

Do instruct models store factual associations differently than base models? 🤔 Doesn't look like it! When adapting ROME's causal tracing code to Llama 3.1 8B, the plots look very similar (base on top, instruct at the bottom). Note the larger sample size for instruct: If the "correct prediction" 1/3

130

Reposted by Neel Rajani

Laura @lauraruis.bsky.social · 20/11/2024

How do LLMs learn to reason from data? Are they ~retrieving the answers from parametric knowledge🦜? In our new preprint, we look at the pretraining data and find evidence against this: Procedural knowledge in pretraining drives LLM reasoning ⚙️🔢 🧵⬇️

36850139

Reposted by Neel Rajani

Ivan Rubachev @puhsu.bsky.social · 25/11/2024

Hello to all #ICLR reviewers on #MLsky

0274

Neel Rajani @neelrajani.bsky.social · 24/11/2024

1/2 The original 2022 ROME paper by Meng et al.:

120