wireless911 / span-aste

a sample pytorch Implementation of ACL 2021 research paper "Learning Span-Level Interactions for Aspect Sentiment Triplet Extraction".
Apache License 2.0
38 stars 2 forks source link

请问bert-large-uncased为什么就训练不动了呢 #15

Open Necro1yte opened 1 year ago

Necro1yte commented 1 year ago

换成了bert-large-uncased,各个metric一直为0

wireless911 commented 1 year ago

可以把你的训练指标贴出来吗

AnitaSherry commented 1 year ago

我也一样,训练指标只是更换了模型的名称,训练了20轮f1才到0.1

wireless911 commented 1 year ago

理论上跟模型无关的,不过这个模型是收敛速度会慢一点,你们可以先使用开源数据来验证模型的结果,然后在你们私有数据上进行超参数调优