More tests

.circleci/config.yml

@@ -6,14 +6,21 @@ version: 2.1
 executors:
   standard_cpu38:
     docker:
-      - image: circleci/python:3.8.0-buster-node
+      - image: cimg/python:3.8.0
     environment:
       PYTHONUNBUFFERED: 1
     resource_class: xlarge
 
   small_cpu38:
     docker:
-      - image: circleci/python:3.8.0-buster-node
+      - image: cimg/python:3.8.0
+    environment:
+      PYTHONUNBUFFERED: 1
+    resource_class: medium
+
+  small_cpu38_conda:
+    docker:
+      - image: continuumio/miniconda3
     environment:
       PYTHONUNBUFFERED: 1
     resource_class: medium
@@ -75,6 +82,27 @@ commands:
             . ~/venv/bin/activate
             python --version
 
+  setup_conda:
+    description: Sets up the virtual environment
+    steps:
+      - run:
+          name: Sets up the virtual environment
+          command: |
+            echo "Started creating the ParlAI conda environment."
+            conda create --quiet --yes --name conda_parlai python=3.8
+            echo "Created the ParlAI conda environment."
+            conda init bash
+            source ~/.bashrc
+            conda deactivate
+            echo "Activating the ParlAI conda environment."
+            conda activate conda_parlai
+            for package in pytorch==1.13.0 torchvision==0.14.0 torchaudio==0.13.0 pytorch-cuda=11.6
+            do
+              echo "Conda installing ${package}."
+              conda install --quiet --yes "${package}" -c pytorch -c nvidia
+            done
+            echo "conda activate conda_parlai" >> $BASH_ENV
+
   codecov:
     description: Coverage report
     steps:
@@ -91,7 +119,8 @@ commands:
       - run:
           name: Installs basic dependencies
           command: |
-            for i in $(seq 1 3); do python -m pip install --progress-bar off pip setuptools==62.3.4 && s=0 && break || s=$? && sleep 10; done; (exit $s)
+            pip install --upgrade pip
+            for i in $(seq 1 3); do python -m pip install --progress-bar off pip setuptools && s=0 && break || s=$? && sleep 10; done; (exit $s)
             for i in $(seq 1 3); do python -m pip install --progress-bar off coverage && s=0 && break || s=$? && sleep 10; done; (exit $s)
             for i in $(seq 1 3); do python -m pip install --progress-bar off codecov && s=0 && break || s=$? && sleep 10; done; (exit $s)
             mkdir -p ~/ParlAI/test-results
@@ -109,7 +138,7 @@ commands:
       - run:
           name: Install torch GPU and dependencies
           command: |
-            python -m pip install --progress-bar off torch==1.11.0+cu113 torchvision==0.12.0+cu113 torchaudio==0.11.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
+            python -m pip install --progress-bar off torch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1
             python -m pip install --progress-bar off 'fairscale~=0.4.0'
             python -m pip install --progress-bar off pytorch-pretrained-bert
             python -m pip install --progress-bar off 'transformers==4.20'
@@ -128,7 +157,7 @@ commands:
           name: Install torch CPU and dependencies
           command: |
             python -m pip install --progress-bar off 'transformers==4.20'
-            python -m pip install --progress-bar off 'torch==1.11.0'
+            python -m pip install --progress-bar off 'torch==1.13.1'
             python -c 'import torch; print("Torch version:", torch.__version__)'
             python -m torch.utils.collect_env
 
@@ -138,7 +167,7 @@ commands:
       - run:
           name: Install torch CPU and dependencies
           command: |
-            python -m pip install --progress-bar off 'torch==1.11.0+cpu' 'torchvision==0.12.0+cpu' 'torchaudio==0.11.0+cpu' -f https://download.pytorch.org/whl/torch_stable.html
+            python -m pip install --progress-bar off 'torch==1.13.1' 'torchvision==0.14.1' 'torchaudio==0.13.1'
             python -m pip install --progress-bar off 'transformers==4.20'
             python -c 'import torch; print("Torch version:", torch.__version__)'
             python -m torch.utils.collect_env
@@ -219,26 +248,26 @@ commands:
             - setupcuda
       - fixgit
       - restore_cache:
-          key: deps-20230209-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
+          key: deps-20230404-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
       - setup
       - installdeps
       - << parameters.more_installs >>
       - save_cache:
-          key: deps-20230209-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
+          key: deps-20230404-<< parameters.cachename >>-{{ checksum "requirements.txt" }}
           paths:
             - "~/venv/bin"
             - "~/venv/lib"
       - findtests:
           marker: << parameters.marker >>
       - restore_cache:
-          key: data-20230209-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
+          key: data-20230404-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
       - run:
           name: Run tests
           no_output_timeout: 60m
           command: |
             coverage run -m pytest -m << parameters.marker >> << parameters.pytest_flags >> --junitxml=test-results/junit.xml
       - save_cache:
-          key: data-20230209-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
+          key: data-20230404-<< parameters.cachename >>-{{ checksum "teststorun.txt" }}
           paths:
             - "~/ParlAI/data"
       - codecov
@@ -255,12 +284,12 @@ commands:
       - checkout
       - fixgit
       - restore_cache:
-          key: deps-20230209-bw-{{ checksum "requirements.txt" }}
+          key: deps-20230404-bw-{{ checksum "requirements.txt" }}
       - setup
       - installdeps
       - installtorchgpu
       - save_cache:
-          key: deps-20230209-bw-{{ checksum "requirements.txt" }}
+          key: deps-20230404-bw-{{ checksum "requirements.txt" }}
           paths:
             - "~/venv/bin"
             - "~/venv/lib"
@@ -301,7 +330,7 @@ commands:
 # -------------------------------------------------------------------------------------
 jobs:
   cleaninstall_38:
-    executor: standard_cpu38
+    executor: small_cpu38
     working_directory: ~/ParlAI
     parallelism: 1
     steps:
@@ -312,7 +341,8 @@ jobs:
           name: Test installation instructions
           no_output_timeout: 60m
           command: |
-            python -m pip install --progress-bar off pip setuptools==62.3.4
+            python -m pip install --progress-bar off --upgrade pip
+            python -m pip install --progress-bar off pip setuptools
             python setup.py develop
             parlai display_data -t integration_tests
 

parlai/crowdsourcing/tasks/dialcrowd/dialcrowd_blueprint.py

@@ -8,7 +8,7 @@
 import logging
 import os
 from dataclasses import dataclass, field
-from typing import Any, Dict, Iterable, TYPE_CHECKING
+from typing import Any, Dict, List, Union, TYPE_CHECKING
 
 from mephisto.operations.registry import register_mephisto_abstraction
 from mephisto.abstractions.blueprint import SharedTaskState
@@ -53,6 +53,8 @@ class DialCrowdStaticBlueprint(StaticReactBlueprint):
     definitions.
     """
 
+    _initialization_data_dicts: Union[List[List[Dict[str, Any]]], List[Dict[str, Any]]]
+
     ArgsClass = DialCrowdStaticBlueprintArgs
     BLUEPRINT_TYPE = STATIC_BLUEPRINT_TYPE
 
@@ -67,7 +69,7 @@ def __init__(
                 f'subtasks_per_unit must be greater than zero but was {self.subtasks_per_unit}'
             )
 
-        self.raw_data: Iterable[Dict[str, Any]] = self._initialization_data_dicts
+        self.raw_data = self._initialization_data_dicts
 
         # Now chunk the data into groups of <num_subtasks>
         grouped_data = []
@@ -77,7 +79,7 @@ def __init__(
         for i in range(0, len(self._initialization_data_dicts), self.subtasks_per_unit):
             chunk = self._initialization_data_dicts[i : i + self.subtasks_per_unit]
             grouped_data.append(chunk)
-        self._initialization_data_dicts = grouped_data
+        self._initialization_data_dicts = grouped_data  # type: ignore
         # Last group may have less unless an exact multiple
         logging.info(
             f'Grouped data into {len(self._initialization_data_dicts)} tasks with {self.subtasks_per_unit} subtasks each.'

parlai/crowdsourcing/tasks/multi_model_chat/__init__.py

@@ -0,0 +1,5 @@
+#!/usr/bin/env python3
+
+# Copyright (c) Facebook, Inc. and its affiliates.
+# This source code is licensed under the MIT license found in the
+# LICENSE file in the root directory of this source tree.

parlai/crowdsourcing/tasks/turn_annotations_static/turn_annotations_blueprint.py

@@ -10,7 +10,7 @@
 import os
 import random
 from dataclasses import dataclass, field
-from typing import Any, Dict, Iterable, List, Optional, TYPE_CHECKING
+from typing import Any, Dict, List, Optional, Union, TYPE_CHECKING
 
 import numpy as np
 from mephisto.operations.registry import register_mephisto_abstraction
@@ -115,6 +115,8 @@ class TurnAnnotationsStaticBlueprint(StaticReactBlueprint):
     definitions.
     """
 
+    _initialization_data_dicts: Union[List[List[Dict[str, Any]]], List[Dict[str, Any]]]
+
     ArgsClass = TurnAnnotationsStaticBlueprintArgs
     BLUEPRINT_TYPE = STATIC_BLUEPRINT_TYPE
 
@@ -133,7 +135,7 @@ def __init__(
                 f'subtasks_per_unit must be greater than zero but was {self.subtasks_per_unit}'
             )
 
-        self.raw_data: Iterable[Dict[str, Any]] = self._initialization_data_dicts
+        self.raw_data = self._initialization_data_dicts
 
         # Load from file if needed specifying which utterances within each
         # conversation to annotate
@@ -180,7 +182,7 @@ def __init__(
         for i in range(0, len(self._initialization_data_dicts), self.subtasks_per_unit):
             chunk = self._initialization_data_dicts[i : i + self.subtasks_per_unit]
             grouped_data.append(chunk)
-        self._initialization_data_dicts = grouped_data
+        self._initialization_data_dicts = grouped_data  # type: ignore
         # Last group may have less unless an exact multiple
         logging.info(
             f'Grouped data into {len(self._initialization_data_dicts)} tasks with {self.subtasks_per_unit} subtasks each.'
@@ -397,7 +399,7 @@ def __init__(
         # (quality control will always be last subtask)
         # TODO: I don't think we need to re-chunk this actually; just iterate
         # over the data and add the quality control task
-        all_data = []
+        all_data: List[Any] = []
         for grp in self._initialization_data_dicts:
             all_data.extend(grp)
 

requirements.txt

@@ -15,8 +15,8 @@ gitdb2==2.0.5
 GitPython==3.0.3
 hydra-core>=1.1.0
 ipython==7.31.1
-torch<1.13.0,>=1.4.0
-torchvision<0.14.0,>=0.5.0
+torch==1.13.1
+torchvision==0.14.1
 joblib==1.2.0
 nltk==3.6.6
 omegaconf>=2.1.1
@@ -44,7 +44,7 @@ subword-nmt==0.3.7
 tensorboardX<=2.5.0
 tokenizers>=0.8.0
 tomli>=2.0.0
-torchtext>=0.5.0,<0.14.0
+torchtext==0.14.1
 tornado==6.0.4
 tqdm~=4.62.1
 typing-extensions==3.7.4.3
@@ -59,4 +59,4 @@ ninja~=1.10.2.3
 protobuf<=3.20.3, >=3.8.0
 contractions~=0.1.72
 fsspec~=2022.2.0
-google-api-core<=2.11.0 # Latest 2.10.2 requires latest protobuf
+google-api-core<=2.11.0

setup.py

@@ -33,7 +33,7 @@
         long_description=readme,
         long_description_content_type='text/markdown',
         url='http://parl.ai/',
-        python_requires='>=3.7',
+        python_requires='>=3.8',
         packages=find_packages(exclude=('data', 'docs', 'tests', 'parlai_internal*')),
         install_requires=reqs,
         include_package_data=True,