wolecki

3 ms·

1. 1.▲
   by wolecki 1y ago

   Some reasoning tokens on this post: >Intelligence too cheap to meter is well within grasp And also: >cost of intelligence should eventually converge to near the cost of electricity. Which is a meter-worthy resource. So intelligence ef
2. 2.▲
   Mind the Trust Gap: Fast, Private Local-to-Cloud LLM Chat(hazyresearch.stanford.edu)

   11 points by wolecki 1y ago|2 comments
3. 3.▲
   by wolecki 1y ago

   A fast Trusted Execution Environment protocol utilizing the H100 confidential mode. Prompts are decrypted and processed in the GPU enclave. Key innovation is that it can be really fast especially on ≥10B parameter models, with latency overh
4. 4.▲
   by wolecki 2y ago

   Paper: https://arxiv.org/abs/2502.15964 Blogpost at Hazy Research: https://hazyresearch.stanford.edu/blog/2025-02-24-minions GitHub repo: https://github.com/HazyResearch/Minio
5. 5.▲
   Minions – hooking up local and cloud LLMs(twitter.com)

   8 points by wolecki 2y ago|2 comments
6. 6.▲
   LoRA Learns Less and Forgets Less(arxiv.org)

   177 points by wolecki 2y ago|60 comments
7. 7.▲
   Math of polygonal billiard tables and illumination(quantamagazine.org)

   2 points by wolecki 3y ago|0 comments