Cognition evaluates trustworthiness of open-source-derived models
HIGHLIGHTS
- •Cognition (Devin) developed an evaluation suite to assess the trustworthiness of open-source-derived models.
- •Ran evaluations on various models, including Kimi K2.7 Code, comparing their SWE-1.7 model to those from leading U.S.-based frontier labs.
- •To determine if their SWE-1.7 model performs as well or better on trustworthiness evaluations compared to other models.
Published:
Official Source (Company Blog)