Spaces:
Running
on
CPU Upgrade
Llama 3 foundation models failing!
Request files (both precisions):
https://huggingface.co/datasets/open-llm-leaderboard/requests/blob/main/meta-llama/Meta-Llama-3-70B_eval_request_False_bfloat16_Original.json
https://huggingface.co/datasets/open-llm-leaderboard/requests/blob/main/meta-llama/Meta-Llama-3-70B_eval_request_False_float16_Original.json
udpate: fp16 has been updated to FINISHED but bf16 is still FAILED
Additionally note that although the FP16 eval is listed as "FINISHED", it does not show up on the leaderboard!
Hi!
This happens sometimes when we have a network failure when pushing results to the hub - tagging
@SaylorTwift
as he'll be able to check on the cluster and push the files (that we always save locally first).
In the meantime, I'm relaunching the bfloat16 version.
This seems to be fixed, thank you!