search

MiuLab / SlotGated-SLU

Slot-Gated Modeling for Joint Slot Filling and Intent Prediction
303 stars 108 forks source link

clarification on paper result #4

Open YuanTingHsieh opened 6 years ago

YuanTingHsieh commented 6 years ago

In your paper, the ATIS Dataset The "Atten Base Liu and Lane 2016" is reported as F1 94.2, Intent Accuracy 91, Sentence Accuracy 78.9

while in their paper and this new paper

is reported as F1 95.98 and Intent Accuracy 98.21

I wonder if your experiment setting is different from theirs? Could you kindly explain this?

Thank you!

zhhongzhi commented 6 years ago

I noticed that this implement used 500 samples from training data as validation set.

sz128 commented 5 years ago

Find solid baselines in https://github.com/sz128/slot_filling_and_intent_detection_of_SLU.