
PhD Université de Montréal & Mila. Currently interning @cohere Interpretability & DL theory
Montréal
Joined October 2024
- Tweets112
- Following164
- Followers178
- Likes133
Pinned Tweet
Latent CoT is an alternative LLM reasoning scheme hypothesized to enable “superposition” allowing models to hold uncertainty over multiple concepts during reasoning 💭
We revisit superposition in 3 latent CoT approaches and find that it is largely an illusion 🔮!
More in 🧵
Michael Rizvi-Martel retweeted
I regret to inform you that regardless of how broken academia is, you are still responsible for what you become during your PhD.
Your advisor can guide you. Your university can give you a degree. Neither can make you an independent thinker.
Stop chasing paper counts. Stop waiting for the perfect research direction. Stop confusing publications with scientific progress.
Stay curious. Question everything. Take ownership of your research.
The goal was never to become a paper-producing machine. It was to become a scientist.
Interesting work by my buddy @KaplFer ! Check it out 👀
Replying to @KaplFer
1/ We pretrain models up to 3B (SmolLM3) from scratch by progressively duplicating layers (Growing) or looping middle layers a fixed number of times (Looping).
We compare both to a full-depth baseline at equal inference FLOPs, with the looped model having fewer parameters.
2
131
Michael Rizvi-Martel retweeted
What would the writing of superintelligent AI sound like? You might expect it to have the characteristics of good human writing — figurative language, high information density, coining new terms, and punchiness — but dialed up to 11. Well, that’s what Claude did, until Anthropic dialed it back recently because of user complaints. It didn’t sound superintelligent to most of us. Instead, the neuralese writing style sounds like a “stupid person’s idea of a smart person”.
Our response to information density and metaphor-laden writing looks something like the cartoon graph I've drawn. Beyond a point, it completely backfires. The reason human writers don’t overdo it isn’t because they aren’t capable, but because they are aware of the reader’s cognitive burden. One of the hardest skills to learn as a writer is to simulate the reader’s experience.
It’s easy to train AI to mimic mediocre human writing. But optimizing it to be (perceived as) good is really hard, and made much harder by the fact that when everyone uses the same few AI models we get tired of their tics very quickly. Most Claudeisms are individually pretty good. I hate that I can’t use “load bearing” in my writing any more. If it weren’t for AI overuse it would be a neat metaphor.
A broader question is whether most tasks in the economy are like chess, where there is a clear performance metric that can be optimized, or whether they’re like writing, where quality is a multidimensional balancing act without clear training signals beyond human mimicry.
Michael Rizvi-Martel retweeted
I was recently interviewed by @CadeMetz for his latest piece on interpretability in the @nytimes 🤖🔍
You find it here 👉 nytimes.com/2026/10/27/scien…
Hoping to see this for future ML conferences! I am happy to review but with a load of 5 papers I can't feasibly give all of them the amount of love they deserve...
Some thoughts on reviewing: spylab.ai/blog/reviewing/
First for @satml_conf where we trialed an early reject phase which cut submissions by ~40%.
Then more generally about how not everything deserves peer-review, and why we should reward writing things that people actually read.
7
329
Michael Rizvi-Martel retweeted
Mila's annual supervision request process is now open to receive MSc and PhD applications for Fall 2027 admission! For more information, visit mila.quebec/en/prospective-s…
Michael Rizvi-Martel retweeted
10 remarks on AI and math that can be true all at once:
1. The recent progress is stunning and exciting.
2. The Quasi-Riemann Hypothesis is arguably Fields-medal level.
3. Mathematics is being transformed and there is no way back.
4. OpenAI doesn't give a fuck about the future of mathematics.
5. Half-baked unreadable preprints claiming major results can effectively stall mathematical progress.
6. Some mathematicians are responding with naive, overly defensive comments.
7. The social-media wankers framing this as "OpenAI tramples gatekeeping" have no experience of mathematical research.
8. The product of mathematics is clarity and understanding. Not theorems, by themselves.
9. Mathematics is a critical infrastructure of human cognition. The human need for understanding isn't obsolete and it can't be offloaded to machines.
10. We haven't "solved" mathematics and we won't get there before the Sun burns out.
Michael Rizvi-Martel retweeted
Cool eval. Simply ask an LLM “Land or Water?” and give it a latitude and longitude coordinate as text. Ask 16,200 times, plot as image. The models know. From compressing the internet.
Michael Rizvi-Martel retweeted
Can Transformer LLMs hide their reasoning from CoT monitors?
tl;dr: Transformers cannot hide complex reasoning, but they might cryptographically obfuscate their CoTs, making it extremely difficult to extract hidden goals.
Michael Rizvi-Martel retweeted
Students sometimes ask me if it still makes sense, in this accelerating age, to pursue a PhD in AI.
Perhaps counterintuitively, I think it's a great time to do so.
I wrote up some thoughts on this here: web.mit.edu/phillipi/www/wri…
Michael Rizvi-Martel retweeted
Check out our new work, to be presented at @COLM_conf!
We find mechanistic evidence that creativity and hallucinations are tied together.
Working on this made me realize that some of the same mechanisms that make LLMs useful may also create failures.
Michael Rizvi-Martel retweeted
Excited that BrainWideBench is finally public! A huge team effort and one of the largest datasets and most comprehensive benchmarks for neural decoders. Glad to have contributed POSSM and to see it do well! Hope the community uses BWB to benchmark their models. 🐭🧠
We’re excited to share BrainWideBench, a new benchmark for evaluating how models of neural activity generalize across animals and tasks!
A major bottleneck in modeling neural data today is rigorous and standardized evaluation. Modelling techniques need to be tested under multiple regimes and downstream tasks.
📄 Check out the paper here! arxiv.org/abs/2609.22064
Special shoutout to @alx_adr, who co-led this work with me and contributed equally as co-first authors!
Thrilled to announce that I will be presenting The Illusion of Superposition at COLM! 🎉
This is my first interp paper and I really enjoyed working on it :)
Thanks again to my amazing co-authors @mariusmosbach and Guillaume Rabusseau
See yall in SF!
Michael Rizvi-Martel retweeted
I will be at #ICML2026 🇰🇷 presenting three papers on interpretability and model adaptation. Feel free to reach out if you want to chat! 🙂
P.S. I'm looking for motivated students to join my new group at Saarland University 🇩🇪: mariusmosbach.com/group/
Michael Rizvi-Martel retweeted
Training a min-max objective, robust/adversarial/self-play training?
Our #ICML2026 paper introduces GOMA: the first method for unconstrained min-max games with a last-iterate stochastic convergence guarantee. 🚀
w/ Jianxin You, @SimonLacosteJ, @ed_gorbunov, @gauthier_gidel.
🧵
Michael Rizvi-Martel retweeted
🚨 Paper Alert! 📣 Speech-language models process text and audio together, but safety is typically evaluated one modality at a time🤔
We show that joint text+audio attacks achieve up to 20× higher jailbreak success rates compared to unimodal attacks😮
If you're into LLMs and looking for a PhD go work with @mariusmosbach! He is a great collaborator and I have no doubt he will make a fine supervisor :)
Replying to @mariusmosbach
I'm looking for a motivated PhD student to start in October 2026, plus research interns, with additional PhD funding coming next year. If you're excited about model adaptation, interpretability, or continual learning, please fill out this form 👉 forms.gle/4r8uvX3uqoPGmKMu7 5/6
Michael Rizvi-Martel retweeted
🚨 I’m happy to share that I'm joining Saarland University as a tenure-track Professor in the Department of Language Science and Technology, and the German Research Center for AI (@DFKI) as a Scientific Director, starting October 2026! 🚨 1/6
Thanks to @che_shr_cat for covering our work! If you want a breakdown of my most recent paper look no further 👀































