vllm feat: quadratic + cubic sampling

This PR adds Quadratic and Cubic Sampling methods to vLLM. Ref: oobabooga/text-generation-webui#5403 and oobabooga/text-generation-webui#5551

Mar 03 '24 16:03 AlpinDale

@AlpinDale Thanks, this looks good to me. Could we add a test?

Mar 04 '24 19:03 Yard1

Thanks for the review @Yard1. It needs a bit more work, could probably use a mask and an unsqueeze for the factor/curve. I'll finish this up soon and add a test unit.

Mar 04 '24 19:03 AlpinDale

@Yard1 Should be done, let me know if there's any issues. Correctness tests pass for the new sampler.

Mar 04 '24 21:03 AlpinDale

Hey @AlpinDale there seems to be an error in the test (the condition is not being met) that was hidden before as logits was modified in place. Could you take a look?

Also the condition itself is very brittle - it would be better to compare the output to some reference.

Mar 05 '24 20:03 Yard1

This pull request has been automatically marked as stale because it has not had any activity within 90 days. It will be automatically closed if no further activity occurs within 30 days. Leave a comment if you feel this pull request should remain open. Thank you!

Oct 30 '24 02:10 github-actions[bot]

This pull request has merge conflicts that must be resolved before it can be merged. @AlpinDale please rebase it. https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

Oct 30 '24 02:10 mergify[bot]

@AlpinDale do you plan to continue this line of work?

Feb 17 '25 18:02 hmellor

@AlpinDale do you plan to continue this line of work?

This sampler is quite outdated by now, so I'm not sure it'll be a useful addition anymore. There are better alternatives.

Feb 18 '25 04:02 AlpinDale

Ok, thank you for the update!

Feb 18 '25 09:02 hmellor