mradermacher/pythia-31m-GGUF
Same weights, converted vs EleutherAI/pythia-31m-deduped. Same weights stored in another dtype (F16/BOOL → F16/F32).
Tensors changed: Embeddings 0/2, Norms 0/26, Attention 0/12, MLP 0/24
Same weights, converted vs EleutherAI/pythia-31m-deduped. Same weights stored in another dtype (F16/BOOL → F16/F32).
Tensors changed: Embeddings 0/2, Norms 0/26, Attention 0/12, MLP 0/24