Baidu releases ERNIE 5.5 Flash Base pretrained model weightsMachine translation

Baidu has released FP8 and BF16 weights for ERNIE 5.5 Flash Base on Hugging Face and ModelScope. The text model has 280 billion parameters in total and activates 14 billion per token. Its weights stop at the end of pretraining, giving teams a common starting point for comparing training methods. In Baidu’s self-reported unified evaluation, it scored highest on 24 of 30 base-model benchmarks but trailed on six.