Dear developers,I currently run a LLAVAGuard-7B on one Nvidia3090 GPU. But I have encountered an extra-long waiting period in loading model weights and the output is "Load weight begin. avail mem=23.33 GB" when it stuck. What shall be the estimated time when loading the model? Also, it seems to be some issue with the use_fast setting in sglang server configuration. the output of which is "Using a slow image processor as use_fast is unset and a slow processor was saved with this model. use_fast=True will be the default behavior in v4.48, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor with use_fast=False".
Thanks for your attention.
Dear developers,I currently run a LLAVAGuard-7B on one Nvidia3090 GPU. But I have encountered an extra-long waiting period in loading model weights and the output is "Load weight begin. avail mem=23.33 GB" when it stuck. What shall be the estimated time when loading the model? Also, it seems to be some issue with the use_fast setting in sglang server configuration. the output of which is "Using a slow image processor as
use_fastis unset and a slow processor was saved with this model.use_fast=Truewill be the default behavior in v4.48, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor withuse_fast=False".Thanks for your attention.