mradermacher/TinyEnsemble-3x1.1B-TinyMoE-GGUF

Same weights, converted vs JoPmt/TinyEnsemble-3x1.1B-TinyMoE. Same weights stored in another dtype (BF16 → F16/F32), plus 66 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/45, Attention 0/88, lm_head 0/1, other 0/22

On the Hub · Downloads on Model Pulse