before this ,I should say I am a totally new of RASA and python. we have a nlu engine base ona rasa nlu 0.13 in before. now we try to upgrade it to new rasa version 2.8 / 3.1.
during the test, I found the current topnology is changed, for some history reason, we already have our own bot but now the rasa have it own bot function in new rasa 2.8/3.1
I downloaded the source code and try to add multi agent in Sonic app framework of current RASA. firstly it seemds sommothly. I have multi agent to hand their own model and could load own model when agents startup.( for example, one for chinse, one for english, one for french etc, with different pipeline component configure and trained before) but after I added the agent number to 5 or 6. there is a warning message has been throw up.
WARNING:tensorflow:5 out of the last 5 calls to <function Model.make_train_function.<locals>.train_function at 0x0000022B2F057E18> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details.
nlu-20220509-162009.tar.gz loaded ./models/nlu-20220509-162009.tar.gz
WARNING:tensorflow:5 out of the last 5 calls to <bound method RasaModel.predict_step of <rasa.nlu.classifiers.diet_classifier.DIET object at 0x0000022B2A87C320>> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for more details.
it seems the tennsoflow components throwed it.
any guys have ideas regarding this? or correspond my requirement, how I could coding for this requirement( mult agent in one Sanic app) ?
thanks a lot