3 ms·
That is for the reinforcement learning part. The base model was likely trained on more GPUs for significantly longer.
by yorwba 1y ago
That is for the reinforcement learning part. The base model was likely trained on more GPUs for significantly longer.