2 ms·
It's unfortunate that the larger model trained to evaluate the technique is "only" 400M parameters, I'd love to see it applied to models with billions of parame
by GaggiX 3y ago
It's unfortunate that the larger model trained to evaluate the technique is "only" 400M parameters, I'd love to see it applied to models with billions of parameters.