Unable to load the model (when not using pipeline)
#8
by
Bhaswanth-RI
- opened
Able to use the pipeline to load and use the model in applications, but unable to load the model directly using model = AutoModel.from_pretrained("sarvamai/shuka-1", trust_remote_code=True)
The instance always crashes everytime I try to load the model (while loading the safetensors). It is not an issue with the gpu memory either, the instance has more than 40gb gpu.
Please help me if anyone knows what the issue here is.