내가 좋아할 만한 콘텐츠
Now you can share your mrr directly from stripe let’s see the real mrrs of companies :D
Ilya says ages of scaling is over and research is back. We need to get ready for new ideas to enhance the AI instead of just compute according to Ilya .
The @ilyasut episode 0:00:00 – Explaining model jaggedness 0:09:39 - Emotions and value functions 0:18:49 – What are we scaling? 0:25:13 – Why humans generalize better than models 0:35:45 – Straight-shotting superintelligence 0:46:47 – SSI’s model will learn from deployment…
Seems like a pretty good model for coding still expensive but much cheaper to its predecessor Opus 4.1. Claude’s almost all focus on currently on coding hence let’s see how it performs on real life tasks .
After the success of training models specifically for a given task, like GPT Codex for coding and Sonnet with its strong focus on coding, there should be more domain specific models like GPT Finance or GPT Math where they excel at a specific task.
Instead of clearly defined problem datasets , these kind of real world engineering benchmarks should be the future for evaluating quality of LLMs. Real world is messy and LLMs should be able to operate in that mess.
We are announcing cline-bench, a real world open source benchmark for agentic coding. cline-bench is built from real world engineering tasks from participating developers where frontier models failed and humans had to step in. Each accepted task becomes a fully reproducible…
This might be biggest bottleneck for training models. Having an efficient data pipeline to train is the biggest moat a company can have.
AI Models are valuable, but datasets and evals to train AI models are more valuable. Datasets are valuable, but automated data pipelines that generate the datasets are more valuable. *** Model < data < pipeline *** At least until the models start building pipelines. Still far…
Big transfer for Thinking Machines. They transferred creator of PyTorch. I hope they also contribute more to open source .
Karpathy explained the new paradigms of the current software era. Verification: if you can verify an output, like passing a unit test or making sure the mathematical output of a model is correct, that is all you need. Real challenge is to find the verifiable domains and specify…
Sharing an interesting recent conversation on AI's impact on the economy. AI has been compared to various historical precedents: electricity, industrial revolution, etc., I think the strongest analogy is that of AI as a new computing paradigm (Software 2.0) because both are…
United States 트렌드
- 1. #StrangerThings5 265K posts
- 2. Thanksgiving 690K posts
- 3. BYERS 61.6K posts
- 4. robin 96.6K posts
- 5. Reed Sheppard 6,299 posts
- 6. Afghan 299K posts
- 7. holly 66.4K posts
- 8. Dustin 89.4K posts
- 9. Podz 4,802 posts
- 10. Vecna 62.2K posts
- 11. Jonathan 75.8K posts
- 12. Erica 18.3K posts
- 13. hopper 16.5K posts
- 14. National Guard 674K posts
- 15. Lucas 84.3K posts
- 16. noah schnapp 9,128 posts
- 17. derek 20K posts
- 18. Joyce 33.5K posts
- 19. Nancy 69.3K posts
- 20. mike wheeler 9,786 posts
Something went wrong.
Something went wrong.