opencompass/CompassJudger-1-14B-Instruct

Full fine-tune vs Qwen/Qwen2.5-14B-Instruct. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Embeddings 1/1, Norms 97/97, MLP 144/144, Attention 336/336, lm_head 1/1

3 derivatives checked

• mradermacher/CompassJudger-1-14B-Instruct-i1-GGUF: Quantization
• featherless-ai-quants/opencompass-CompassJudger-1-14B-Instruct-GGUF: Quantization
• mradermacher/CompassJudger-1-14B-Instruct-GGUF: Quantization

On the Hub · Downloads on Model Pulse