seamon67/Wald-4B

Full fine-tune (fewer layers) vs org2ai/Wald-4B. seamon67/Wald-4B matches part of org2ai/Wald-4B (426 of its 738 tensors; the rest of org2ai/Wald-4B has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 1/1, Norms 105/105, Attention 201/224, MLP 96/96

On the Hub · Downloads on Model Pulse