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
Our org does not allow network download of file on our deployment servers, so we need to uplaod the file to a fixed read only directory on the server. Being able to configure the path for that server would be useful.
Describe alternatives you've considered
I tried setting the TMPDIR env variable, but that works at python global level for all libraries and does not seem very configurable.
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
Currently Google aiplatform loads tokenizer by either downloading from github or reading from a tmpdir cache path which is not configurable at a library level. https://github.com/googleapis/python-aiplatform/blob/main/vertexai/tokenization/_tokenizer_loading.py#L136-L147
Can we make it configurable like how TikToken or NLK does it ? https://github.com/openai/tiktoken/blob/main/tiktoken/load.py#L34-L42 With a env variable like
VERTEX_TOKENIZER_CACHE_DIR
?Describe the solution you'd like
Our org does not allow network download of file on our deployment servers, so we need to uplaod the file to a fixed read only directory on the server. Being able to configure the path for that server would be useful.
Describe alternatives you've considered
I tried setting the TMPDIR env variable, but that works at python global level for all libraries and does not seem very configurable.
The text was updated successfully, but these errors were encountered: