rizki4106 / netrade Goto Github PK

torchvision package is missing in readme.md for installation instructions

torchvision version 0.13.1

AttributeError: module 'torchvision.transforms' has no attribute 'ToTenser'

on

chart_transformer = transforms.Compose([
    transforms.Resize((224,224)),
    transforms.ToTensor()
])

after preparing my own images in the file structure I ran this section of code. It crashes my computer every time.

from netrade.data import DataPreprocessing

# initialize class
chart = "neTrade/chart/"
candle = "neTrade/candle/"

prep = DataPreprocessing(chart_path=chart, candle_path=candle)

# create dataframe
frame = prep.create_frame()

After running this about 30 seconds go by and then it soft-locks my computer. It could be that maybe I prepared too many images. I ran tree | wc -l to tell me how many total images I have. I have 30755 images split evenly between charts and candles.

Any help diagnosing this issue would be very helpful.

Traceback (most recent call last):
File "C:/Users/bill/Desktop/teemaki/ai netrade/ai.py", line 90, in
model, history = netrade.train(X_train=dataloader,X_test=dataloader , epochs=10)
File "C:\Users\bill\AppData\Local\Programs\Python\Python310\lib\site-packages\netrade\core.py", line 57, in train
accuracy_score = Accuracy().to(device)
TypeError: Accuracy.new() missing 1 required positional argument: 'task'

So far while trying out this library I have had a great time. Since this library relies so heavily on the images generated; you should include a code snippet of how you generate your Charts and Candles. With this improvement, anyone could pick up the library and use yfinance for example to train their own model very easily