Cognition's SWE-1.7 strips CCP bias from a Chinese open-source base model and matches US frontier labs on trustworthiness evals, at a fraction of the cost.
Full article content could not be extracted automatically. Read the original below.
Cognition's SWE-1.7 Strips Chinese AI Bias From Open-Source Code Models
Cognition's SWE-1.7 strips CCP bias from a Chinese open-source base model and matches US frontier labs on trustworthiness evals, at a fraction of the cost.
Topics: