# Latest

**URL:** https://discourse.edwardlib.org/latest.md

[Latest](https://discourse.edwardlib.org/latest.md) · [Categories](https://discourse.edwardlib.org/categories.md)

---

## [Factor graph + belief prop model in Edward?](https://discourse.edwardlib.org/t/factor-graph-belief-prop-model-in-edward/548)

<div class="topic-metadata">

**Author:** [@BoltzmannBrain](https://discourse.edwardlib.org/u/BoltzmannBrain)\
**Replies:** 1\
**Last updated:** [March 1, 2022, 12:34pm UTC](https://discourse.edwardlib.org/t/factor-graph-belief-prop-model-in-edward/548 "2022-03-01T12:34:11Z")

</div>

Hi all, I’m looking for some direction or examples on implementing a factor graph + belief propagation model in Edward. For example, I have a factor graph representation of sequential states (variables) that have Gauss…

---

## [Getting started](https://discourse.edwardlib.org/t/getting-started/943)

<div class="topic-metadata">

**Author:** [@RossChapman](https://discourse.edwardlib.org/u/RossChapman)\
**Replies:** 5\
**Last updated:** [June 9, 2020, 9:16pm UTC](https://discourse.edwardlib.org/t/getting-started/943 "2020-06-09T21:16:38Z")

</div>

Hi all I am having problems with the example code from http://edwardlib.org/getting-started. The code runs to the line: inference.run(n\_samples=5, n\_iter=250)\_ where it stops with the error: IndexError: list index o…

---

## [Tensorflow errror](https://discourse.edwardlib.org/t/tensorflow-errror/1114)

<div class="topic-metadata">

**Author:** [@saurabh1088](https://discourse.edwardlib.org/u/saurabh1088)\
**Replies:** 0\
**Last updated:** [June 9, 2020, 9:12pm UTC](https://discourse.edwardlib.org/t/tensorflow-errror/1114 "2020-06-09T21:12:00Z")

</div>

I am trying to use a simple example from Edwards on TensorFlow and I have tried both on jupyter notebook and on pycharm, the code breaks in the inference section run.inference() and when I trace it back it is breaking in…

---

## [Using Point Processes in Edward](https://discourse.edwardlib.org/t/using-point-processes-in-edward/421)

<div class="topic-metadata">

**Author:** [@jaivardhankapoor](https://discourse.edwardlib.org/u/jaivardhankapoor)\
**Replies:** 2\
**Last updated:** [April 16, 2020, 2:26pm UTC](https://discourse.edwardlib.org/t/using-point-processes-in-edward/421 "2020-04-16T14:26:20Z")

</div>

Hi all, I am working on a project that requires modeling of events using a Hawkes Process(an inhomogenous point process) along with a heirarchical graphical model. Since there is no support for point processes in Edward…

---

## [Discrete Bayesian Network inference with Edward](https://discourse.edwardlib.org/t/discrete-bayesian-network-inference-with-edward/1106)

<div class="topic-metadata">

**Author:** [@fahimeh](https://discourse.edwardlib.org/u/fahimeh)\
**Replies:** 0\
**Last updated:** [February 27, 2020, 5:05pm UTC](https://discourse.edwardlib.org/t/discrete-bayesian-network-inference-with-edward/1106 "2020-02-27T17:05:00Z")

</div>

Hi, My goal is to use Edward library for discrete data where each node in the network has CPT instead of PDF and applying the Bayesian Network inference. The following link aims to demo how to use Edward for discrete …

---

## [Should pip install edward\>=2.0.0 just use github.com/google/edward2?](https://discourse.edwardlib.org/t/should-pip-install-edward-2-0-0-just-use-github-com-google-edward2/1086)

<div class="topic-metadata">

**Author:** [@dustin](https://discourse.edwardlib.org/u/dustin)\
**Replies:** 1\
**Last updated:** [February 24, 2020, 3:03pm UTC](https://discourse.edwardlib.org/t/should-pip-install-edward-2-0-0-just-use-github-com-google-edward2/1086 "2020-02-24T15:03:22Z")

</div>

GitHub issue: https://github.com/blei-lab/edward/issues/937 We’re setting up a standalone repository for Edward2 (https://github.com/google/edward2). This will let us further develop the excellent examples from Edward1 …

---

## [Edward essay: Probabilistic Programming Possibilities](https://discourse.edwardlib.org/t/edward-essay-probabilistic-programming-possibilities/1102)

<div class="topic-metadata">

**Author:** [@NicT](https://discourse.edwardlib.org/u/NicT)\
**Replies:** 0\
**Last updated:** [January 29, 2020, 4:53pm UTC](https://discourse.edwardlib.org/t/edward-essay-probabilistic-programming-possibilities/1102 "2020-01-29T16:53:28Z")

</div>

Hello, I am new to this forum (and sort of new to Edward as well). I wrote an essay based on reviewing a few academic papers serving as an introduction to the library including a demonstration of Variational Auto-Encoder…

---

## [Black box likelihood](https://discourse.edwardlib.org/t/black-box-likelihood/1096)

<div class="topic-metadata">

**Author:** [@zacmon](https://discourse.edwardlib.org/u/zacmon)\
**Replies:** 0\
**Last updated:** [December 6, 2019, 8:52pm UTC](https://discourse.edwardlib.org/t/black-box-likelihood/1096 "2019-12-06T20:52:59Z")

</div>

Hi, I currently use PYMC3 to use NUTS on a likelihood function and it’s gradients that I have a C++ function calculate. (I can’t use tensorflow because my calculations require arbitrary precision. Tensorflow’s arbitrary…

---

## [Black box variational inference with no model derivatives](https://discourse.edwardlib.org/t/black-box-variational-inference-with-no-model-derivatives/1091)

<div class="topic-metadata">

**Author:** [@mukesh\_r](https://discourse.edwardlib.org/u/mukesh_r)\
**Replies:** 0\
**Last updated:** [October 1, 2019, 2:50am UTC](https://discourse.edwardlib.org/t/black-box-variational-inference-with-no-model-derivatives/1091 "2019-10-01T02:50:26Z")

</div>

Does Edward have Black Box Variational Inference using Score Gradient? I have a Finite Element Code as the model, therefore I cannot evaluate the model derivatives (implies, I cannot have likelihood derivatives). So I w…

---

## [Conditional probabilities from a neural network](https://discourse.edwardlib.org/t/conditional-probabilities-from-a-neural-network/1085)

<div class="topic-metadata">

**Author:** [@RossChapman](https://discourse.edwardlib.org/u/RossChapman)\
**Replies:** 0\
**Last updated:** [August 12, 2019, 6:18am UTC](https://discourse.edwardlib.org/t/conditional-probabilities-from-a-neural-network/1085 "2019-08-12T06:18:11Z")

</div>

Hi all, I have built a neural network that predicts the values of a continuous variable using using Edward. I can make predictions about the expected value of the output variable eg: res=np.vstack(\[sess.run(y\_post,{x\_…

---

## [Re-using a saved Bayesian neural network](https://discourse.edwardlib.org/t/re-using-a-saved-bayesian-neural-network/997)

<div class="topic-metadata">

**Author:** [@RossChapman](https://discourse.edwardlib.org/u/RossChapman)\
**Replies:** 2\
**Last updated:** [August 7, 2019, 12:58am UTC](https://discourse.edwardlib.org/t/re-using-a-saved-bayesian-neural-network/997 "2019-08-07T00:58:05Z")

</div>

Hi all, I have a question about how to re-use Bayesian neural networks built with Edward. I have built and trained a simple Bayesian neural network. I imagine that I might want to re-use this model for analyses some t…

---

## [Example of Implicitqp with local variables](https://discourse.edwardlib.org/t/example-of-implicitqp-with-local-variables/1080)

<div class="topic-metadata">

**Author:** [@www3cam](https://discourse.edwardlib.org/u/www3cam)\
**Replies:** 0\
**Last updated:** [July 29, 2019, 11:19pm UTC](https://discourse.edwardlib.org/t/example-of-implicitqp-with-local-variables/1080 "2019-07-29T23:19:11Z")

</div>

Hi, I’m trying to train a nonlinear statespace model with implicitqp function and I’m wondering if there are examples with local variables being used. The example on the Edward github is a bayesian linear regression wi…

---

## [Implementing ADVI](https://discourse.edwardlib.org/t/implementing-advi/1079)

<div class="topic-metadata">

**Author:** [@jomg](https://discourse.edwardlib.org/u/jomg)\
**Replies:** 0\
**Last updated:** [July 27, 2019, 1:37am UTC](https://discourse.edwardlib.org/t/implementing-advi/1079 "2019-07-27T01:37:49Z")

</div>

Hi everyone, I’ve implemented ADVI (i think) using ReparameterizationKLqp class, in order to approximate the posterior of a Gamma variable, Z. To do so, I define the input latent variables dictionary as {Z : LogNormal} …

---

## [Reproducing CVAE in Edward 2 and Keras](https://discourse.edwardlib.org/t/reproducing-cvae-in-edward-2-and-keras/1074)

<div class="topic-metadata">

**Author:** [@ahmadsalim](https://discourse.edwardlib.org/u/ahmadsalim)\
**Replies:** 0\
**Last updated:** [July 3, 2019, 6:07pm UTC](https://discourse.edwardlib.org/t/reproducing-cvae-in-edward-2-and-keras/1074 "2019-07-03T18:07:59Z")

</div>

Hi, I am trying to reproduce the Convolutional VAE tutorial from Tensorflow (https://www.tensorflow.org/beta/tutorials/generative/cvae), by using Keras and Edward’s facilities for probabilistic programming. (Partially i…

---

## [Help for Beginner using Edward](https://discourse.edwardlib.org/t/help-for-beginner-using-edward/1064)

<div class="topic-metadata">

**Author:** [@samoalfred](https://discourse.edwardlib.org/u/samoalfred)\
**Replies:** 1\
**Last updated:** [July 2, 2019, 6:36pm UTC](https://discourse.edwardlib.org/t/help-for-beginner-using-edward/1064 "2019-07-02T18:36:45Z")

</div>

I am new to using Edward. I am trying to do the tutorials but I have noticed that two kinds of errors exist despite importing the correct libraries. ImportError: cannot import name ‘set\_shapes\_for\_outputs’ ImportError:…

---

## [Epistemic uncertainity using Edwards](https://discourse.edwardlib.org/t/epistemic-uncertainity-using-edwards/829)

<div class="topic-metadata">

**Author:** [@CSharmi](https://discourse.edwardlib.org/u/CSharmi)\
**Replies:** 1\
**Last updated:** [May 9, 2019, 9:15am UTC](https://discourse.edwardlib.org/t/epistemic-uncertainity-using-edwards/829 "2019-05-09T09:15:29Z")

</div>

Hello, I am learning how to use Edward following the tutorials which is an excellent knowledge repository. I am trying to build a simple two layer neural network following the tutorials. I am wondering if I could estima…

---

## [Practical experiences using ADVI training Bayesian neural network](https://discourse.edwardlib.org/t/practical-experiences-using-advi-training-bayesian-neural-network/836)

<div class="topic-metadata">

**Author:** [@pswpswpsw](https://discourse.edwardlib.org/u/pswpswpsw)\
**Replies:** 5\
**Last updated:** [May 9, 2019, 9:00am UTC](https://discourse.edwardlib.org/t/practical-experiences-using-advi-training-bayesian-neural-network/836 "2019-05-09T09:00:05Z")

</div>

Hello! I am just simply using KLqp, and my neural network model is very simple so it would be using the reparameteric gradient, which is the ADVI algorithm in the ADVI paper. I have found that using Adam optimizer is…

---

## [Edward for categorical multi class classification](https://discourse.edwardlib.org/t/edward-for-categorical-multi-class-classification/1023)

<div class="topic-metadata">

**Author:** [@latent](https://discourse.edwardlib.org/u/latent)\
**Replies:** 1\
**Last updated:** [May 9, 2019, 8:32am UTC](https://discourse.edwardlib.org/t/edward-for-categorical-multi-class-classification/1023 "2019-05-09T08:32:22Z")

</div>

Is Edward support classification for categorical (using any encoding method such as one hot encoding) multi class classification? if so , which of the tutorial can guide me through that ? Thanks

---

## [InvalidArgumentError while running conditioning on Edward](https://discourse.edwardlib.org/t/invalidargumenterror-while-running-conditioning-on-edward/1054)

<div class="topic-metadata">

**Author:** [@harc007](https://discourse.edwardlib.org/u/harc007)\
**Replies:** 0\
**Last updated:** [April 5, 2019, 9:35am UTC](https://discourse.edwardlib.org/t/invalidargumenterror-while-running-conditioning-on-edward/1054 "2019-04-05T09:35:55Z")

</div>

Hi, I’m new to Edward and getting stuck at this issue. import edward as ed import six import tensorflow as tf from edward.models import Bernoulli, Beta unknown = Beta(concentration1=1.0, concentration0=1.0) c = Bernoull…

---

## [Can we use bayesian NN for multi class classification](https://discourse.edwardlib.org/t/can-we-use-bayesian-nn-for-multi-class-classification/1049)

<div class="topic-metadata">

**Author:** [@rajeevanarun](https://discourse.edwardlib.org/u/rajeevanarun)\
**Replies:** 0\
**Last updated:** [March 16, 2019, 9:27pm UTC](https://discourse.edwardlib.org/t/can-we-use-bayesian-nn-for-multi-class-classification/1049 "2019-03-16T21:27:41Z")

</div>

Can we use bayesian NN for multi class classification

---

## [RNN in Edward - Tensor had NaN values](https://discourse.edwardlib.org/t/rnn-in-edward-tensor-had-nan-values/719)

<div class="topic-metadata">

**Author:** [@chdr](https://discourse.edwardlib.org/u/chdr)\
**Replies:** 2\
**Last updated:** [March 13, 2019, 9:05am UTC](https://discourse.edwardlib.org/t/rnn-in-edward-tensor-had-nan-values/719 "2019-03-13T09:05:45Z")

</div>

I am new to Edward and relatively new to TensorFlow. I try to build a Bayesian RNN to predict remaining useful life on NASA’s CMAPSS dataset. I started from the RNN example: Edward RNN and tried to get it running. imp…

---

## [Bayesian RNN in Edward](https://discourse.edwardlib.org/t/bayesian-rnn-in-edward/1047)

<div class="topic-metadata">

**Author:** [@abdullahnisar92](https://discourse.edwardlib.org/u/abdullahnisar92)\
**Replies:** 0\
**Last updated:** [March 13, 2019, 8:34am UTC](https://discourse.edwardlib.org/t/bayesian-rnn-in-edward/1047 "2019-03-13T08:34:37Z")

</div>

Hello Everybody Does anybody has idea whether it is possible to run inference on Bayesian LSTM or Bayesian Recurrent Neural Network model in Edward.? BR Abdullah

---

## [Categorical distribution support](https://discourse.edwardlib.org/t/categorical-distribution-support/1045)

<div class="topic-metadata">

**Author:** [@Srikanth](https://discourse.edwardlib.org/u/Srikanth)\
**Replies:** 0\
**Last updated:** [March 4, 2019, 3:22am UTC](https://discourse.edwardlib.org/t/categorical-distribution-support/1045 "2019-03-04T03:22:50Z")

</div>

How to create a categorical distribution that has support of -1,0,1 instead of 0,1,2?

---

## [Acceptance Rate 0 for pareto/nbd model](https://discourse.edwardlib.org/t/acceptance-rate-0-for-pareto-nbd-model/1026)

<div class="topic-metadata">

**Author:** [@nresnick](https://discourse.edwardlib.org/u/nresnick)\
**Replies:** 0\
**Last updated:** [January 20, 2019, 11:09pm UTC](https://discourse.edwardlib.org/t/acceptance-rate-0-for-pareto-nbd-model/1026 "2019-01-20T23:09:55Z")

</div>

Here’s the full, reproducible code. I’m banging my head against my desk trying to figure out why the HMC sampler isn’t working. Any help would be greatly appreciated! import tensorflow as tf import tensorflow\_probabilit…

---

## [Product of distributions as variational family](https://discourse.edwardlib.org/t/product-of-distributions-as-variational-family/1040)

<div class="topic-metadata">

**Author:** [@senqi](https://discourse.edwardlib.org/u/senqi)\
**Replies:** 0\
**Last updated:** [February 11, 2019, 9:51pm UTC](https://discourse.edwardlib.org/t/product-of-distributions-as-variational-family/1040 "2019-02-11T21:51:55Z")

</div>

I’m a beginner at Edward. I wonder if there’s a way to define the variational family of a random variable as the product of, say, unit Gaussian distributions with the same mean and variance? For example, how can I defin…

---

## [Categorical Variational Inference](https://discourse.edwardlib.org/t/categorical-variational-inference/1038)

<div class="topic-metadata">

**Author:** [@Srikanth](https://discourse.edwardlib.org/u/Srikanth)\
**Replies:** 0\
**Last updated:** [February 7, 2019, 9:18pm UTC](https://discourse.edwardlib.org/t/categorical-variational-inference/1038 "2019-02-07T21:18:44Z")

</div>

How to modify the Categorical distribution to obtain a support over -1,0,1 and not the default 0,1,2 for a three event experiment. Basically I want to apply variational inference by choosing the categorical distribution …

---

## [Why is there no mention of TensorFlowProbability/Edward2 on the edwardlib website?](https://discourse.edwardlib.org/t/why-is-there-no-mention-of-tensorflowprobability-edward2-on-the-edwardlib-website/1034)

<div class="topic-metadata">

**Author:** [@dmerrell](https://discourse.edwardlib.org/u/dmerrell)\
**Replies:** 0\
**Last updated:** [January 30, 2019, 11:16pm UTC](https://discourse.edwardlib.org/t/why-is-there-no-mention-of-tensorflowprobability-edward2-on-the-edwardlib-website/1034 "2019-01-30T23:16:31Z")

</div>

Could somebody please indicate on the edwardlib website that Edward is not being maintained anymore, and that newcomers should look at TensorFlow Probability/Edward2 instead? I was about to dive into learning Edward whe…

---

## [Help with a simple parameter estimation](https://discourse.edwardlib.org/t/help-with-a-simple-parameter-estimation/1029)

<div class="topic-metadata">

**Author:** [@quertenmont](https://discourse.edwardlib.org/u/quertenmont)\
**Replies:** 0\
**Last updated:** [January 22, 2019, 6:22am UTC](https://discourse.edwardlib.org/t/help-with-a-simple-parameter-estimation/1029 "2019-01-22T06:22:37Z")

</div>

Hello, I have a problem to solve, where I am trying to constraint posterior based on prior distribution and relations between my random variables. I am starting in edward so I wanted to test a very simple case: 3 rand…

---

## [How to set up HMC with custom log probability](https://discourse.edwardlib.org/t/how-to-set-up-hmc-with-custom-log-probability/1024)

<div class="topic-metadata">

**Author:** [@nresnick](https://discourse.edwardlib.org/u/nresnick)\
**Replies:** 0\
**Last updated:** [January 17, 2019, 4:02pm UTC](https://discourse.edwardlib.org/t/how-to-set-up-hmc-with-custom-log-probability/1024 "2019-01-17T16:02:11Z")

</div>

I’m trying to replicate this pymc3 model in edward2 (model is in the “PyMC implementation” section). I have this set up so far: # Given our current guess for lamda and mu, compute the log likelihood at x, tx, T def pnb…

---

## [Acceptance Rate 0 for HMC in IRT models](https://discourse.edwardlib.org/t/acceptance-rate-0-for-hmc-in-irt-models/999)

<div class="topic-metadata">

**Author:** [@AkmusNie](https://discourse.edwardlib.org/u/AkmusNie)\
**Replies:** 1\
**Last updated:** [January 18, 2019, 3:14am UTC](https://discourse.edwardlib.org/t/acceptance-rate-0-for-hmc-in-irt-models/999 "2019-01-18T03:14:01Z")

</div>

I am running HMC to learn a so called item response model that is common in my practice to estimate student abilities and question difficulties. With only these two groups of parameters, the algorithm seems to work ok, b…

[Next page](https://discourse.edwardlib.org/latest.md?page=1)