projects

Project summaries from atifquamar.com.

research
project thumbnail

Adaptive Blockwise Search: Inference-Time Alignment for Large Language Models

Adaptively focuses computation on the most critical early tokens during LLM decoding, boosting alignment performance across multiple tasks compared to Best-of-N and fine-tuning.
project thumbnail

Learning Modal-Mixed Chain-of-Thought Reasoning with Latent Embeddings

Modal-mixed chain-of-thought lets a VLM interleave text with compact latent visual sketches, using a diffusion-based latent decoder with SFT+RL training to boost vision-intensive reasoning while adding only modest inference overhead.
project thumbnail

STARS - Segment-level Token Alignment via Rejection Sampling in Large Language Models

Decoding method that aligns large language models with human preferences at inference time by accepting only high-reward text segments, boosting quality without retraining.