yoshino0721/Xiangxue-1-3.6B

Partial fine-tune vs twnlp/ChineseErrorCorrector4-4B. Partial change: 89% of the weights changed (attention, mlp, norms), in 32 of 32 layers.

Tensors changed: Embeddings 0/1, Norms 120/129, MLP 96/96, Attention 128/128

1 derivatives checked

• mradermacher/Xiangxue-1-3.6B-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse