
Hi, I'm Mohsen!
and I love math ❤️. I work on these things.
I'm always excited to start new collaborations, especially when it's something new to me. Feel free to reach out by email; you'll find it in the footer.
News
All news
Medical Image Spatial Grounding with Semantic Sampling
MICCAI 2026 poster: MIS-Ground tests 3D anatomical spatial grounding across CT, MRI, views, coordinates, prompts, and terminology. MIS-SemSam changes the token-selection rule and raises Qwen3-VL-32B accuracy from 53.4% to 66.5%.
Quantize What Counts: More for Keys, Less for Values
A geometry-driven mixed-precision KV-cache quantization poster showing that keys carry more information than values, so key-favored bit allocation preserves accuracy while reducing memory.
Ranking Reasoning LLMs under Test-Time Scaling
Ranking reasoning LLMs under repeated sampling, comparing 72 ranking methods across four Olympiad-style math benchmarks and packaging them in Scorio.
Don't Pass@k: A Bayesian Framework for Large Language Model Evaluation
Proposed a Bayesian framework that estimates models' success probabilities with quantified uncertainty, yielding more reliable rankings and enabling categorical evaluation of LLMs.
70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float (DFloat11)
NeurIPS 2025 poster on DFloat11: a lossless compression framework that shrinks LLMs and diffusion transformers to ~70% of their size with bit-for-bit identical outputs, plus a GPU kernel that decompresses on the fly.
Serving Reasoning LLMs Efficiently and Reliably [No Anime]
These slides use figures from four papers to explain DFloat11 weight compression, KV-cache quantization, and how repeated sampling changes reasoning-model rankings.
Serving Reasoning LLMs Efficiently and Reliably
Animated examples show how DFloat11 and KV-cache quantization save GPU memory, then follow repeated answers through Bayes@N evaluation and model ranking.
Python Environments
Python environments, how to create and reproduce them, and when to use pip, conda, micromamba, uv, pipx, lockfiles, and containers.