When I insert some AI generated text input, perplexity returns infinity. Because of this issue, if else condition failing. Do you know best solution to handle this issue? Thank you
@AIAnytime7 ай бұрын
Can you create a GitHub issue? I will try to reproduce.
@niharikasingh99637 ай бұрын
can you please help me out after running it it is showing this error AttributeError: module 'torch.nn' has no attribute 'cross_entropy'
@adityarajpateriya3870 Жыл бұрын
I tried this model and there are very few instances where it said , "text is generated by AI",most of the text are classified as human-generated, so can you suggest me some enhancements, or can i adjust values of perplexity and bustiness to get the desired result ? i need to submit it as college project
@AIAnytime Жыл бұрын
Hi Aditya, Your observation is right! Can you check the threshold value and see if there is any improvement?
@Itsme-zb3zr7 ай бұрын
Hi, How you have finalised this perplexity and business limit In which value this will give the best result? I'm getting "text not generated by AI" for all AI generated texts. What adjustment needs to be done to fix this
@Armagedonnable7 ай бұрын
Did you figure out issue? I face same problem@@Itsme-zb3zr
@Armagedonnable7 ай бұрын
Same here@@Itsme-zb3zr
@priyalpavanipriyal56136 ай бұрын
@@Itsme-zb3zr I'm facing the same issue ig the more the model improves more the threshold values keep varying so idk If u can solve this issue let me know
@SD_Web_Wizrads Жыл бұрын
Good website and good work sir
@AIAnytime Жыл бұрын
Thank you, Suhas!
@SwatiSingh-b4r5 ай бұрын
Can you make a video on plagiarism checker application using python too..
@minatouzumaki446810 ай бұрын
bro i am always getting : [nltk_data] Error loading punckt: Package 'punckt' not found in index [nltk_data] Downloading package stopwords to [nltk_data] C:\ProgramData\anaconda3\lib ltk_data... [nltk_data] Package stopwords is already up-to-date! Downloading model.safetensors: 0%|
@Brian-nh8qk Жыл бұрын
can you like do for gpt 4 please
@anirudhcodes Жыл бұрын
Awesome 👌
@hammadhussain49575 ай бұрын
please give complete code of this, your github repo has different code and doesnt work
@bec_Divyansh Жыл бұрын
Why this difference between burstiness and perplexity? why is perplexity dependent on model type but not burstiness?
@AIAnytime Жыл бұрын
Hi Divyansh. Burstiness refers to the uneven distribution of word occurrences in the corpus. Perplexity not only depends on model types but also prediction tasks. Burstiness can affect language modeling in many ways like data preprocessing, vocabulary size, contextual dependencies, unseen data perform, etc etc. Perplexity is more concerned with evaluating the model's predictive accuracy at the local level. Burstiness on other hand, encompasses broader statistical patterns in the word distribution. It can impact overall characteristics and is difficult to calculate.