JAX array conversion failure in Keras model prediction about keras HOT 3 CLOSED

jax.errors.TracerArrayConversionError: The numpy.ndarray conversion method array() was called on traced array with shape float32[10000,2].

This indicates that your model contains an operation that tries to retrieve the eager value of a tensor.

Earlier, I see:

jnp.exp(mass_model.log_prob(m1q) + self.logVT.predict(m1m2).flatten()),

So it sounds like you are calling predict() inside a tracing scope. This is impossible. Perhaps you meant to call self.logVT(m1m2) instead?

See also: https://keras.io/getting_started/faq/#whats-the-difference-between-model-methods-predict-and-call

from keras.

Oh, and if your call method is stateful in any way, you'll need to use stateless_call() instead and manage the state updates manually.

from keras.

So it sounds like you are calling predict() inside a tracing scope. This is impossible. Perhaps you meant to call self.logVT(m1m2) instead?

@fchollet This solved the problem. Thanks!

from keras.