Harry Potter GPT -- Base (pretrain-only) stage

Part of a from-scratch pretrain -> SFT -> DPO pipeline. This checkpoint is the Base (pretrain-only) stage.

Full write-up: https://abhijitdalal.vercel.app/projects/harry-potter-gpt Source: https://github.com/abhijitdalal26/harry-potter-gpt

Downloads last month
32
Safetensors
Model size
0.1B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support