search

cvlab-columbia / viper

Code for the paper "ViperGPT: Visual Inference via Python Execution for Reasoning"
Other
1.65k stars 115 forks source link

Disk quota for running main_batch.py #30

Closed Anthony6197 closed 9 months ago

Anthony6197 commented 1 year ago

Hi,

When I run the main_batch with CONFIG_NAME=base_config python main_batch.py, It downloads files that saturated my disk quota 20G (I am working on a remote cluster) and report disk quota exceed error and then exit. May I ask if that is what quota should I expect by running main_batch with base_configs (I only modify the directory of pre-trained model and data)? Here is a screen shot of the error:

Screenshot 2023-06-17 at 7 20 05 PM Screenshot 2023-06-17 at 7 20 27 PM
surisdi commented 1 year ago

Hi, the vision models are very large. Just downloading the BLIP with flan-t5-xxl will take more space than the 20GB that you have available. It would be ideal to get a larger disk quota. Otherwise, you can try using smaller models (e.g. change the blip_v2_model_type: blip2-flan-t5-xxl configuration for another pre-trained BLIP one).

surisdi commented 9 months ago

Marking the stale issue as resolved. Feel free to reopen if there are any updates.