mradermacher/OpenELM-1_1B-instruct-fix-nan-GGUF

Same weights, converted vs jasonkrone/OpenELM-1_1B-instruct-fix-nan. Same weights stored in another dtype (BF16 → F16/F32), plus 56 new tensors.

Tensors changed: Norms 0/113, Attention 0/56, MLP 0/28, Embeddings 0/1

On the Hub · Downloads on Model Pulse