ZERO-POINT-AI/undermaintenance__118b-gpt-faulty

Partial fine-tune vs openai/gpt-oss-120b. Partial change: 30% of the weights changed (attention, lm_head, mlp), in 36 of 36 layers.

Tensors changed: MLP 5/144, Norms 0/73, Attention 67/324, lm_head 1/1, Embeddings 0/1

On the Hub · Downloads on Model Pulse