| ▲ | selcuka 6 hours ago | |
It's ok if they never release a BF16 model, but it's less ok if they release it, win the benchmarks, then quantise it after a few weeks. | ||
| ▲ | retinaros 2 hours ago | parent [-] | |
that is for sure what everyone does. also they train on evals with the datasets that they would be bench against. | ||