Views
No views yet
transformers-compatible model. To load it, download gpt_best.pt
and use the same GPT class definition from the training notebook:1ckpt = torch.load("gpt_best.pt", map_location="cpu", weights_only=False)
2model = GPT(ckpt["config"])
3model.load_state_dict(ckpt["model_state_dict"])
4stoi, itos = ckpt["stoi"], ckpt["itos"]