dangell7/Condor-27B

Full fine-tune vs Jackrong/Qwopus3.5-27B-v3. Fine-tune: 95% of the weights changed, across 64 of 64 layers.

Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 164/209, Attention 409/448, MLP 192/192

1 derivatives checked

On the Hub · Downloads on Model Pulse