Open smurf-1119 opened 7 months ago
You can substitute the cfg.TRAINER.GRAPHADAPTER.BETA as a fixed value, such as 0.2, 0.3, 0.4. When I am free, I will refine the basic code for the convenient setting with the script, Thanks.
Notably that the setting of beta will be important for the adapter-style tuning.
Thanks for your comments.
I encounter
AttributeError: GRAPHADAPTER
when I run the scriptscripts/graphclip/main_v1.sh
.Traceback (most recent call last): File "train.py", line 213, in
main(args)
File "train.py", line 147, in main
trainer = build_trainer(cfg)
File "/Dassl.pytorch/dassl/engine/build.py", line 11, in build_trainer
return TRAINER_REGISTRY.get(cfg.TRAINER.NAME)(cfg)
File "/Dassl.pytorch/dassl/engine/trainer.py", line 325, in init
self.build_model()
File "/GraphAdapter/trainers/baseclip_graph_v1.py", line 354, in build_model
self.model = CustomCLIP(cfg, classnames, clip_model, self.train_loader_x).cuda()
File "/GraphAdapter/trainers/baseclip_graph_v1.py", line 313, in init
self.graph_learner = GraphLearner(cfg, classnames, clip_model, base_text_features, base_img_features)
File "/GraphAdapter/trainers/baseclip_graph_v1.py", line 191, in init
self.beta_it = cfg.TRAINER.GRAPHADAPTER.BETA
File "/home/miniconda3/envs/dassl/lib/python3.8/site-packages/yacs/config.py", line 141, in getattr
raise AttributeError(name)
AttributeError: GRAPHADAPTER
Could you please tell me how to fix it?