# Jul 26, 2026

## Today

### Daily

### Weekly

- Running a 28.9M parameter LLM on an $8 microcontroller

- A **28.9 million parameter language model** runs on the **ESP32-S3 microcontroller**, leveraging **Per-Layer Embeddings** to store most parameters in flash memory, allowing it to operate efficiently despite limited SRAM.

- The new rules of context engineering for Claude 5 generation models

- **Claude 5 models** have undergone significant changes in context engineering, notably reducing the system prompt by over **80%** without sacrificing performance, allowing for more flexible and intuitive interactions.

- Bringing PyTorch Monarch to AMD GPUs

- **PyTorch Monarch** has been successfully ported to **AMD GPUs** with ROCm, enabling **fault-tolerant distributed training** that minimizes wasted computation and maximizes GPU utilization, crucial for training large language models.

- Need Arxiv endorsement for CS.LG to publish a preprint about Learning Stable Latent Manifolds from noisy sensor data /Observation space \[P\]

- The study explores **learning stable latent manifolds** from **noisy sensor data**, demonstrating that such data can effectively support auto-regression tasks under specific conditions, with empirical validation on NASA's CMAPSS dataset.