Bump pyro-ppl requirement to >=1.8.4 (#1606)

Summary:
This includes two major changes of interest to botorch:
1. Support for pytorch 2.0 (allowing us to not require pyro dev branch when testing against pytorch nightlies): pyro-ppl/pyro#3164
2. The ability to registering of custom exception handlers for numerical issues: pyro-ppl/pyro#3168

Pull Request resolved: #1606

Reviewed By: saitcakmak

Differential Revision: D42330914

Pulled By: Balandat

fbshipit-source-id: 1117e7713b99819b4f1297de2d49aeb1bd2b9c9b

Author: Balandat

.conda/meta.yaml

@@ -22,7 +22,7 @@ requirements:
     - linear_operator ==0.2.0
     - scipy
     - multipledispatch
-    - pyro-ppl >=1.8.2
+    - pyro-ppl >=1.8.4
 
 test:
   imports:

.github/workflows/deploy_on_release.yml

@@ -86,7 +86,7 @@ jobs:
         conda install -y scipy sphinx pytest flake8 multipledispatch
         conda install -y -c pytorch pytorch cpuonly
         conda install -y -c gpytorch gpytorch
-        conda install -y -c conda-forge pyro-ppl>=1.8.2
+        conda install -y -c conda-forge pyro-ppl>=1.8.4
         conda config --set anaconda_upload no
     - name: Build and verify conda package
       shell: bash -l {0}

.github/workflows/nightly.yml

@@ -105,7 +105,7 @@ jobs:
         conda install -y scipy multipledispatch setuptools_scm conda-build conda-verify
         conda config --set anaconda_upload no
         conda install -y -c pytorch-nightly pytorch cpuonly
-        conda install -y -c conda-forge pyro-ppl>=1.8.2
+        conda install -y -c conda-forge pyro-ppl>=1.8.4
         pip install git+https://github.com/cornellius-gp/linear_operator.git
         pip install git+https://github.com/cornellius-gp/gpytorch.git
     - name: Build and verify conda package

.github/workflows/test_stable.yml

@@ -48,7 +48,7 @@ jobs:
         conda install -y -c pytorch pytorch cpuonly
         conda install -y pip scipy pytest
         conda install -y -c gpytorch gpytorch
-        conda install -y -c conda-forge pyro-ppl>=1.8.2
+        conda install -y -c conda-forge pyro-ppl>=1.8.4
         pip install .[test]
     - name: Unit tests
       shell: bash -l {0}

README.md

@@ -59,7 +59,7 @@ Optimization simply use Ax.
 - PyTorch >= 1.11
 - gpytorch == 1.9.0
 - linear_operator == 0.2.0
-- pyro-ppl >= 1.8.2
+- pyro-ppl >= 1.8.4
 - scipy
 - multiple-dispatch
 

docs/getting_started.md

@@ -19,7 +19,7 @@ Before jumping the gun, we recommend you start with the high-level
 - linear_operator == 0.2.0
 - scipy
 - multiple-dispatch
-- pyro-ppl >= 1.8.2
+- pyro-ppl >= 1.8.4
 
 BoTorch is easily installed via
 [Anaconda](https://www.anaconda.com/distribution/#download-section) (recommended)

environment.yml

@@ -9,4 +9,4 @@ dependencies:
   - linear_operator==0.2.0
   - scipy
   - multipledispatch
-  - pyro-ppl>=1.8.2
+  - pyro-ppl>=1.8.4

requirements.txt

@@ -1,6 +1,6 @@
 multipledispatch
 scipy
 torch>=1.11,<2.0
-pyro-ppl>=1.8.2
+pyro-ppl>=1.8.4
 gpytorch==1.9.0
 linear_operator==0.2.0