Datasets:

PatronusAI

/

requests

like 0

Sure, the results should be updated in some time.

@sunitha-ravi It failed again, but my 7B params model ran successfully.

Pallas-0.5 (the model that failed) is 34B. Do you support 34B models? I noticed that all the models in the leaderboard are 13B or smaller.

Yes we do! Our backend space unexpectedly went to sleep in the middle of the run, I will trigger a restart!

@sunitha-ravi Thank you!

I'm glad you do support 34b. The quality jump from 7b models is quite high, imo.

I'm looking forward to the retrigger (nothing new appears in the history of https://huggingface.co/datasets/PatronusAI/requests/blob/main/Mihaiii/Pallas-0.5_eval_request_False_bfloat16_Original.json ).

We have had a lot of requests and our backend was down temporarily, it should be on the board soon!