Math solving pure reinforcement learning Is there a case…
0 reactions
Math solving pure reinforcement learning
Is there a case where a math solver or theorem prover or even just an ai that does basic addition trained solely on reinforcement learning?
I have seen some but they use a base model or base llm and post train it or fine tune, but I am asking for a pure rl approach.
#technology
Is there a case where a math solver or theorem prover or even just an ai that does basic addition trained solely on reinforcement learning?
I have seen some but they use a base model or base llm and post train it or fine tune, but I am asking for a pure rl approach.
#technology
| 6,000 mlx | |
| 1 | |
| 0 |
comments
Pure RL from zero for math sounds hard, most setups still sneak in base model first actually.
