Reeves289/Spark-X2.5-4B

Full fine-tune vs XHToken/Spark-X2.5-4B-Base. Fine-tune: 100% of the weights changed, across 36 of 36 layers.

Tensors changed: Embeddings 1/1, Norms 73/73, MLP 108/108, Attention 108/108

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

On the Hub · Downloads on Model Pulse