mamelles/LFM2.5-8B-A1B-Wolof-CPT-v3

Partial fine-tune vs LiquidAI/LFM2.5-8B-A1B-Base. Partial change: 0% of the weights changed (attention, mlp), in 22 of 24 layers.

Tensors changed: MLP 22/2162, Embeddings 0/1, other 0/54, Norms 0/60, Attention 18/24

On the Hub · Downloads on Model Pulse