All Projects Khmer TTS DecisionOps Inference Engineering Embedding Diagnostics Enigmatica XAI Vision Transfer Customer Feedback Intelligence SSL and Semi-supervised Learning on MNIST

All posts · 57

Latest

Beyond text: what transfers to vision, speech and multimodal models

Speech AI · 8 posts

Newest

LLM inference · 7 posts

Newest

Representation learning · 1 post

01Zero loss, empty representation

Foundations · 15 posts

Start with

Model development · 26 posts

Start with

All Paper Reviews

Technical reports

Tülu 3: an open post-training recipe and what its ablations show

Papers

Chinchilla: how much data a model should see, and how firmly the paper shows it DeepSeekMath and GRPO: a cheaper RL algorithm, and what RL actually changed DPO: preference learning without a reward model, and how strong the evidence was FlashAttention: exact attention made fast by counting memory traffic All 16 papers

Contact

Open to Data Scientist and ML/AI Engineer opportunities, collaborations, and technical discussions.