Benchmark for LLaMA pruning

[dercaft]

If it is possible to give a benchmark of Pruned LLaMA.

[horseee]

Hi.

It's on our waitlist, but it requires a large amount of time and resources to conduct post-training on the pruned model. Otherwise, the pruned model is poorly functioning.

We are still trying to find a possible and efficient solution for post-training the pruned model since the dataset is not public. If you have any ideas, we would be grateful for any suggestions or contributions you may have.

[horseee]

We have updated the evaluation results in https://github.com/horseee/LLM-Pruner. Please refer to the new repo-v-