Closed AEDWIP closed 6 years ago
Hi Andy, If you want to use the up-to-date version of Chiron please download the clean-code branch. git checkout clean-code should do the job.
There is also a new model in this branch, which should work with Tensorflow 1.3.0. But for other versions of Tensorflow, you have to train your model using that version of Tensorflow.
Another way is to freeze the model using the chiron/export_test.py, and run the froze model with chiron/load_run_test.py, but these scripts are still under development.
Hi Lee
It looks like my virtualenv is using the correct version of tensorflow. It looks like there a lot of new development going on. Maybe I'll wait till clean-code branch gets merged back. I would prefer to use python3 and a newer version of tensorflow.
(Chiron_env) $ python --version Python 2.7.10 (Chiron_env) $ python -c 'import tensorflow as tf; print tf.version' 1.0.1 (Chiron_env) $
Kind regards
Andy
Hi Haotianteg
I am happy to train a new model. Is your training data set available?
Kind regards
Andy
Clean_code branch is not ready, please wait when it merge to master.
Hi Andy, I have made a patch in the clean_code branch, just tested it on Tensorflow 1.3 Cuda8 and Tensorflow 1.5 Cuda9, all work fine now. If there is still problem you can continue reply under this thread.
Welcome for contributing, especially on training new models.
Thanks Teng
Thanks
andy
From: Lee Yam Keng notifications@github.com Reply-To: haotianteng/Chiron <reply+00725a2fd0c699f20ca15a9bad472db6b7216ae96434b14092cf0000000116c66ab89 2a169ce12217c2e@reply.github.com> Date: Sunday, March 18, 2018 at 10:58 AM To: haotianteng/Chiron Chiron@noreply.github.com Cc: Andrew Davidson Andy@SantaCruzIntegration.com, Author author@noreply.github.com Subject: Re: [haotianteng/Chiron] unable to run 'chiron/entry.py call' (#30)
And Teng will send the download link to get the full dataset for training.
I am getting a bunch of errors when I try to call using the sample data. I wonder if the pretrained model is missing?
Kind regards
Andy
generates a lot of messages like
eventually I get the following stack trace
entry.py.call.error.out.txt