Mostafa8Mehrabi/qwen3-71M-c4-final

Full fine-tune vs Mostafa8Mehrabi/qwen3-50m-fp16. Fine-tune: 100% of the weights changed, across 8 of 8 layers.

Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 33/33, MLP 24/24, Attention 32/32

1 derivatives checked

• ItsVictorTube/Qwen3-Neurotic-Experts-248x71M: MoE built from dense models

On the Hub · Downloads on Model Pulse