February 2026

I’ve been experimenting with reinforcement learning and built a small project that trains a PPO agent to play Super Mario Bros locally. Mostly did it to better understand SB3 and training dynamics instead of just running example notebooks. It uses a Gym-compatible NES environment + Stable-Baselines3 (PPO). I added a simple FastAPI server that streams frames to a browser UI so I can watch the agent during training instead of only checking TensorBoard. What I’ve been focusing on: […]

Ver mais

Personal AI agents incur substantial cost via repeated LLM calls. We show existing caching methods fail: GPTCache achieves 37.9% accuracy on real benchmarks; APC achieves 0-12%. The root cause is optimizing for the wrong property — cache effectiveness requires key consistency and precision, not classification accuracy. We observe cache-key evaluation reduces to clustering evaluation and apply V-measure decomposition to separate these on n=8,682 points across MASSIVE, BANKING77, CLINC150, and NyayaBench v2, our new 8,514-entry multilingual agentic dataset (528 […]

Ver mais

I’ve written before about computing big Fibonacci numbers, and about creating a certificate to verify a Fibonacci number has been calculated correctly. This post will revisit both, giving a different approach to computing big Fibonacci numbers that produces a certificate along the way. As I’ve said before, I’m not aware of any practical reason to compute large Fibonacci numbers. However, the process illustrates techniques that are practical for other problems. The fastest way to compute the nth Fibonacci number […]

Ver mais

Camera recapture introduces complex optical degradations, such as perspective warping, illumination shifts, and Moiré interference, that remain challenging for deep watermarking systems. We present TIACam, a text-anchored invariant feature learning framework with auto-augmentation for camera-robust zero-watermarking. The method integrates three key innovations: (1) a learnable auto-augmentor that discovers camera-like distortions through differentiable geometric, photometric, and Moiré operators; (2) a text-anchored invariant feature learner that enforces semantic consistency via cross-modal adversarial alignment between image and text; and (3) a […]

Ver mais

Optimally trading-off exploration and exploitation is the holy grail of reinforcement learning as it promises maximal data-efficiency for solving any task. Bayes-optimal agents achieve this, but obtaining the belief-state and performing planning are both typically intractable. Although deep learning methods can greatly help in scaling this computation, existing methods are still costly to train. To accelerate this, this paper proposes a variational framework for learning and planning in Bayes-adaptive Markov decision processes that coalesces variational belief learning, sequential […]

Ver mais