Downloads · 30 days
0
gpt2
triton329/gpt2
gpt2 is a machine learning model from triton329. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
GPT-style language model built from scratch in PyTorch, trained on a subset of WikiText-103 corpus. Model trained on an NVIDIA A100 GPU — full run completes in ~6 minutes.
Access
Public
Updated Jun 22, 2026
Repo size
2.7 GB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.pth498 MB · 100%
At a glance
- Access
- Public
- Created
- Jun 21, 2026
- Updated
- Jun 22, 2026
- SHA
- 0b550e19
Model card
From the Hugging Face model README
GPT-2 in PyTorch
GPT-style language model built from scratch in PyTorch, trained on a subset of WikiText-103 corpus. Model trained on an NVIDIA A100 GPU — full run completes in ~6 minutes.
Install
python -m venv venv && venv\Scripts\activate
pip install -r requirements.txt
Train
python main.py --gpu --tokenizer tiktoken
Inference
python inference.py --prompt "Once upon a time" --gpu
Pretrained weights load automatically from triton329/gpt2 on HuggingFace.
Example output
Once upon a time . At the early 20th century , though it was seen by the time
in 1864 , it lacks a 1804 Antigua – 5 @-@ class submarine of the hurricane...
- Created
- Jun 21, 2026
- Updated
- Jun 22, 2026