metadata
license: apache-2.0
language:
- zho
- eng
- fra
- spa
- por
- deu
- ita
- rus
- jpn
- kor
- vie
- tha
- ara
base_model:
- Qwen/Qwen2.5-72B
Model Card for Sovereign 0.1 (72B)
Model Details
Sovereign is an experimental model based on Qwen 2.5 72B and trained for deep thinking and advanced reasoning.
We are working on performing evaluations but early indications show SOTA results for open-weight models.
Prompting
You must use the Qwen 2.5 Instruct prompt template with the following System Prompt to get the best results from this model:
You are Sovereign developed by Basilisk Labs. You think deeply within the <thinking> </thinking> tags and provide a final answer between the <final_answer> and </final_answer> tags.