search

luheng / deep_srl

Code and pre-trained model for: Deep Semantic Role Labeling: What Works and What's Next
Apache License 2.0
330 stars 77 forks source link

Model taking more than 4 hours to run #23

Open mario-rggj opened 5 years ago

mario-rggj commented 5 years ago

I am receiving this error on ubuntu machine

Model building and loading duration was 0:00:28.

Running model duration was 3:24:52. Task: srl Allow new words in test data: True Embedding size=100 Read 295330 sentences. Data loading duration was 0:00:30. Using 2 feature types, projected output dim=200. ('lstm_0_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faa7c08d110> ('lstm_1_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faa7c08d7d0> ('lstm_2_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faac68df910> ('lstm_3_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faac68df350> ('lstm_4_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faac68dfcd0> ('lstm_5_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faac68f7c50> ('lstm_6_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faac68f76d0> ('lstm_7_rdrop', 0.1, True) <neural_srl.theano.layer.HighwayLSTMLayer object at 0x7faac68f7b90> 46989 32054 46989 2 Loaded model from: /home/ubuntu/deep_srl-master/resources/conll05_ensemble/model1.npz Model building and loading duration was 0:00:03.

and one model is taking about 4 hours to run ... any idea what to doo?

_Originally posted by @Nasidiqi in https://github.com/luheng/deep_srl/issues/21#issuecomment-484228925_

mario-rggj commented 5 years ago

@nasidiqi Not sure if this is an issue, maybe its about your computational resources avaiable what machine are you running this on?

try running the conll05_model instead of the ensemble

Nasidiqi commented 5 years ago

I am using ubuntu 18.4. Can you please check this thread

Error while running endtoend.sh and its taking too long on big dataset #22

it has complete trace of error