at2005

4 ms·

1. 1.▲
   Matrix Orthogonalization Improves Memory in Recurrent Models(ayushtambde.com)

   85 points by at2005 3mo ago|32 comments
2. 2.▲
   by at2005 7mo ago

   I didn't compare with the harness (focused on distillation) but the original ToT paper has a section on it: https://arxiv.org/abs/2305.10601
3. 3.▲
   by at2005 7mo ago

   Ah, I meant that MCTS uses more inference-time compute (over GRPO) to produce a training sample
4. 4.▲
   Tree Search Distillation for Language Models Using PPO(ayushtambde.com)

   87 points by at2005 7mo ago|10 comments
5. 5.▲
   by at2005 6y ago

   Btw the whole motivation for this were algorithms like Grover's, which need "oracles" to be specified. You can only imagine trying to code adders and greater-than circuits with QASM...
6. 6.▲
   A HL Programming Language for Quantum Computers

   1 points by at2005 6y ago|1 comments