faridbagheri/qwen3-0.6b-masakhapos-yoruba-full-ft
Full fine-tune vs Qwen/Qwen3-0.6B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 68/113, MLP 84/84, Attention 112/112
Full fine-tune vs Qwen/Qwen3-0.6B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 68/113, MLP 84/84, Attention 112/112