mradermacher/pyGPT-50M-GGUF
Same weights, converted vs 0x7o/pyGPT-50M. Same weights stored in another dtype (F32/U8 → F16/F32).
Tensors changed: lm_head 0/1, Attention 0/16, Norms 0/34, MLP 0/16, Embeddings 0/2
Same weights, converted vs 0x7o/pyGPT-50M. Same weights stored in another dtype (F32/U8 → F16/F32).
Tensors changed: lm_head 0/1, Attention 0/16, Norms 0/34, MLP 0/16, Embeddings 0/2