mradermacher/gpt-neox-20b-GGUF
Quantization vs EleutherAI/gpt-neox-20b. mradermacher/gpt-neox-20b-GGUF is a GGUF quantization (Q8_0) of the same weights: 310 tensors match exactly and 134 within quantization error.
Tensors changed: Embeddings 0/2, Attention 0/88, Norms 0/178, MLP 0/176