BBQGOD/DeepSeek-GRM-16B

Full fine-tune vs deepseek-ai/DeepSeek-V2-Lite. Fine-tune: 100% of the weights changed, across 27 of 27 layers.

Tensors changed: lm_head 1/1, Embeddings 1/1, Norms 82/82, MLP 2808/2808, Attention 108/108

2 derivatives checked

• mradermacher/DeepSeek-GRM-16B-i1-GGUF: Quantization
• mradermacher/DeepSeek-GRM-16B-GGUF: Quantization

On the Hub · Downloads on Model Pulse