Search
Meta Faces Challenges with Llama 4 Model: DeepSeek Counterattack Fails, Performance Lags
2025-04-08 /
Read about 0 minute
Author：小编
In a social media post at 1:30 AM today, Ahmad Al-Dahle, the head of Meta's generative AI division, addressed questions raised about the recently open-sourced Llama 4 model. Ahmad clarified that Llama 4 was hastily released following its development, leading to inconsistencies in model quality across various services. Meta has vowed to swiftly address vulnerabilities and improve performance. Furthermore, he denied any claims that Llama 4 underwent pre-training on test sets.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
