ServiceNow-AI/Apriel-5B-Instruct

Full fine-tune vs ServiceNow-AI/Apriel-5B-Base. Fine-tune: 100% of the weights changed, across 28 of 28 layers.

Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 57/57, MLP 84/84, Attention 112/112

1 derivatives checked

• mrfakename/Apriel-5B-Instruct-llamafied: Identical copy

2 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse