You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I am using the latest huggingface transformers (version==4.48.0.dev0). When I tried to run the demo from here, I have this error: AttributeError: 'LlamaForCausalLM' object has no attribute 'set_kv_cache_offload'.
Does anyone know the solution to solve this issue? Thanks!
Activity
kdw4537 commentedon Mar 21, 2025
I'm having the same problem (OPT as well)
Did you build a tokenizer in fork of transformers? In my case, failed.
tjruwase commentedon Mar 21, 2025
@yuzhenmao, sorry that we missed this issue. The kv cache offload was prototyped in a transformer's fork, but never upstreamted.
https://github.com/deepspeedai/DeepSpeedExamples/blob/master/inference/huggingface/zero_inference/README.md#install
@kdw4537, are you interested in this feature?
Freed-Wu commentedon Jun 12, 2025
It cannot be installed. @tjruwase