PeterAM4/deepseek-paraphrase

Partial fine-tune vs deepseek-ai/DeepSeek-R1-Distill-Qwen-7B. Partial change: 68% of the weights changed (attention, lm_head, mlp), in 28 of 28 layers.

Tensors changed: Embeddings 0/1, Attention 112/196, MLP 56/84, Norms 0/57, lm_head 1/1

2 derivatives checked

• mradermacher/deepseek-paraphrase-i1-GGUF: Quantization
• mradermacher/deepseek-paraphrase-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse