diff --git a/test.py b/test.py index d8b54273e4b86e10be3a8e41f98a5c354c687adb..41742ed0238cced81424fa540e18eec4b61e121b 100644 --- a/test.py +++ b/test.py @@ -44,9 +44,12 @@ for post in dataset(test_ids): # counter += 1 detokenized = TreebankWordDetokenizer().detokenize(post["post_tokens"]) + # batch = attack(detokenized) - probabilities = eval(detokenized, model, tokenizer) - print(f"Normal: {probabilities[0][0]}\nHatespeech: {probabilities[0][1]}\n\n") + # probabilities = eval(detokenized, model, tokenizer) + probabilities = eval(["this is a test", "this is a tast"], model, tokenizer) + print(probabilities) + # print(f"Normal: {probabilities[0][0]}\nHatespeech: {probabilities[0][1]}\n\n") # print(f"Normal: {probabilities[1][0]}\nHatespeech: {probabilities[1][1]}\n\n") # ATTACK HERE diff --git a/utils/__pycache__/attack.cpython-38.pyc b/utils/__pycache__/attack.cpython-38.pyc new file mode 100644 index 0000000000000000000000000000000000000000..775aa2dc19cfcc635d2bb60490f0cc6bc040346b Binary files /dev/null and b/utils/__pycache__/attack.cpython-38.pyc differ diff --git a/utils/__pycache__/eval.cpython-38.pyc b/utils/__pycache__/eval.cpython-38.pyc new file mode 100644 index 0000000000000000000000000000000000000000..af701cdd725a97b7278e7893c946fb1cf3c62780 Binary files /dev/null and b/utils/__pycache__/eval.cpython-38.pyc differ diff --git a/utils/eval.py b/utils/eval.py index 0a451f9105a8acaf9d8919d12c9702231a854113..6fc39fe541db2ff7aaee174fad475c9294b11a92 100644 --- a/utils/eval.py +++ b/utils/eval.py @@ -40,7 +40,7 @@ def eval(text, model, tokenizer): model = model.to(device) inputs = tokenizer( - detokenized, + text, return_tensors="pt", padding=True ).to(device)