mergekit-community/BetterGPT2

Same weights, converted vs openai-community/gpt2-xl. Same weights stored in another dtype (F32 → F16).

Tensors changed: Attention 0/192, Norms 0/194, MLP 0/192, Embeddings 0/2

2 derivatives checked

• mradermacher/BetterGPT2-GGUF: Same weights, converted
• mergekit-community/OpenGPT-3: Identical copy

2 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse