gpt-2-simple
gpt-2-simple copied to clipboard
Python package to easily retrain OpenAI's GPT-2 text-generating model on new texts
I’m wondering if there is anyway to that I can display gpt2’s output word by word instead of waiting for it to complete generating everything and then printing?
I have realised that load_dataset.py fails if there is an empty string exist in loaded CSV file. That causes script to give the following error: raw_text += start_token + row[0]...
How to train with GPU on my own pc? Can anyone give a brief introduction?
I have a dataset of **6.6 GB** size and I am using the `355M` model. I realized from this [comment](https://github.com/minimaxir/gpt-2-simple/issues/53#issuecomment-493718734) that the dataset should be smaller than or equal the...
Hi, So i'm trying to finetune the medium model on a TITAN RTX (24G) in WSL2 but it seems to run out of memory? the small model fits. If i...
Verify that the largest model is supported: * Finetuning is disabled * Generating from default is supported * Update docs.
**Problem Statement :** npz files do not load with Newer numpy versions > 1.16 **The Fix :** This is done by editing the load_dataset.py file by specifying the allow_pickle parameter...
Closes #221
Hi all - amazing repo - small issue: I can't supply a file path outside of the root directory in my gdrive to `copy_file_from_gdrive` Code is here : https://github.com/minimaxir/gpt-2-simple/blob/master/gpt_2_simple/gpt_2.py#L618 ```...
Hello: Thanks for the repo and the numerous updates. Have everything working fine except reading lines from a text file for 'gpt2.generate' function. Would like to read line-by-line instead of...