qingy2024/Qwen2.5-4B

Partial fine-tune vs Qwen/Qwen2.5-3B. Partial change: 75% of the weights changed (attention, mlp, norms), in 30 of 36 layers.

Tensors changed: Embeddings 0/1, Norms 60/73, MLP 90/108, Attention 210/252

1 derivatives checked

• DevQuasar-7/qingy2024.Qwen2.5-4B-GGUF: Quantization

On the Hub · Downloads on Model Pulse