Open Deepayan137 opened 6 years ago
Hey, you've mentioned a satisfactory flag in the code and while the flag remains untrue the loop should go on. My question is when does the flag turn to true. Won't the present loop be infinite?
full codecode
from ocr.pytorch.engine import Engine satisfactory = False savepath = "file.tar" kwargs = {} try: with open(savepath, "rb") as savefile: kwargs = torch.load(savefile) except FileNotFoundError: kwargs = { "input_size": 32, "output_classes": len(lmap.keys()), "lr": 3e-4, "momentum": 0.8 } #train_set = train_set[:300] #validation_set = validation_set[:30] engine = Engine(**kwargs) while not satisfactory: val_err, train_err = engine.train(train_set, validation_set, debug=True) kwargs = engine.export() torch.save(kwargs, open(savepath, "wb+"))
This is from memory. Satisfactory condition is somewhere below the while block. It's a validation loss of 35ish or something.
Hey, you've mentioned a satisfactory flag in the code and while the flag remains untrue the loop should go on. My question is when does the flag turn to true. Won't the present loop be infinite?
full codecode