Server message:Endpoint failed to start

Inference Endpoints on the Hub
1

Hello,

My model was working and suddenly it did not work anymore. I keep getting this error:

"Server message:Endpoint failed to start. Back-off pulling image “registry.internal.huggingface.tech/hf-endpoints/inference-pytorch-gpu:sha-00395f5"”

Tried restarting, and recreating endpoints many times but it persists.

I’m running “amazon/chronos-t5-large” with custom handler functions for embeddings on Nvidia T4.

The endpoint is protected.

I will highly appreciate your help as this matter is critical for me.

1 Like
2

Same thing happened to me

3

For me pressing the “Update endpoint” and then pausing and unpausing the endpoint seems to have resolved the issue.