| user: | smaddrellmander |
| created: | April 21, 2026 |
| karma: | 92 |
| about: | idlemachines.co.uk |
| 1. | Thinking Machines dropped RoPE, and it's a good idea(idlemachines.co.uk) |
| 2. | Nesso-1: Accelerating Open-Source Binding Affinity Predictions [pdf](valencelabs.com) |
| 3. | The annotated PyTorch training loop(idlemachines.co.uk) |
| 4. | Mae vs. MSE: more than just the mean vs. median debate(idlemachines.co.uk) |
| 5. | DiffusionGemma: Discrete diffusion in a large language model(idlemachines.co.uk) |
| 6. | Heaven knows I'm perplexed now(idlemachines.co.uk) |
| 7. | Reading MAI's efficiency gain. How to pick architectures like serious people(idlemachines.co.uk) |
| 8. | MAI-Thinking-1: Building a Hill-Climbing Machine [pdf](microsoft.ai) |
| 9. | Are contrastive losses just cross entropy all along?(idlemachines.co.uk) |
| 10. | Every token, everywhere, all at once(idlemachines.co.uk) |
| 11. | The cut in the Mixture of Experts compute graph(idlemachines.co.uk) |
| 12. | DeepSeek V4 from the Inside(idlemachines.co.uk) |
| 13. | Softmax, can you derive the Jacobian? And should you care?(idlemachines.co.uk) |
| 14. | Gemma 4 is not your standard transformer(idlemachines.co.uk) |