mrtoots/unsloth-gpt-oss-20b-mlx-fp16

Same weights, converted vs unsloth/gpt-oss-20b. Same weights stored in another dtype (U8/BF16 → F16), plus 144 new tensors.

Tensors changed: Norms 0/49, MLP 0/48, Attention 0/216, lm_head 0/1, Embeddings 0/1

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse