mradermacher/megatron-gpt2-345m-GGUF
Same weights, converted vs robowaifudev/megatron-gpt2-345m. Same weights stored in another dtype (F16/U8 → F16/F32).
Tensors changed: Attention 0/48, Norms 0/98, MLP 0/48, Embeddings 0/2
Same weights, converted vs robowaifudev/megatron-gpt2-345m. Same weights stored in another dtype (F16/U8 → F16/F32).
Tensors changed: Attention 0/48, Norms 0/98, MLP 0/48, Embeddings 0/2