The week ahead

Every Monday, I publish a brief note that captures my top of mind. These could include:

  • Highlights - wins and milestones worth celebrating.
  • Sigals - trends or data points that might matter later.
  • Finds - miscellaneous reads I stumble upon worth sharing.

The week ahead: 20251201

[Blog] HOPE. A model as a set of nested sub-optimizers, each with its own context flow and hyperparams. Relevant paper.

The week ahead: 20251020

[Paper] Synthetic customers for purchase intent prediction. Rather than generating a Likert score for intent (and there would be many relevant techniques ...

The week ahead: 20251013

[Paper] Embedding initialization shown to be related to Grokking (delayed generalization). On the XOR task, embeddings trained on a small model is shown t...

The week ahead: 20251006

[Blog] Thinking Machines Lab: LoRA is better to be applied to all weights, including MLP and MoE. On RL, LoRA performs on par with full training even with...

The week ahead: 20250929

[Announcement] ChatGPT Pulse. ChatGPT can now proactively push personalized content.

The week ahead: 20250915

[Blog] First glimpse into the work of Thinking Machines Lab. Inference systems can be made deterministic by using “batch-invariant” GPU kernels (GitHub im...

The week ahead: 20250811

[HuggingFace] OpenAI’s open-weight model. Technical report from OpenAI. Onboarding guide from HuggingFace.

The week ahead: 20250804

[Blog] Utkarsh’s insights on the mathematical impossibility of some expectations of AI agents in 2025. TLDR: error compounding, token economics, tool engi...

The week ahead: 20250721

[Blog] Perplexity to launch its own browser, with rumors of OpenAI’s upcoming release of a similar product, signaling challenges to Google Chrome’s contro...