Bump docker images to torch 2.3.1 (#3366)

* Revert "Autoresume Validation with Max Duration (#3358)" This reverts commit f0eae8a. * bump images to 2.3.1 * bump torchvision
mosaicml · Jun 5, 2024 · 86dfaf7 · 86dfaf7
1 parent 74a5f78
commit 86dfaf7
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 24 deletions.
diff --git a/docker/README.md b/docker/README.md
@@ -30,9 +30,9 @@ To install composer, once inside the image, run `pip install mosaicml`.
 <!-- BEGIN_PYTORCH_BUILD_MATRIX -->
 | Linux Distro   | Flavor   | PyTorch Version   | CUDA Version        | Python Version   | Docker Tags                                                                              |
 |----------------|----------|-------------------|---------------------|------------------|------------------------------------------------------------------------------------------|
-| Ubuntu 20.04   | Base     | 2.3.0             | 12.1.1 (Infiniband) | 3.11             | `mosaicml/pytorch:latest`, `mosaicml/pytorch:2.3.0_cu121-python3.11-ubuntu20.04`         |
-| Ubuntu 20.04   | Base     | 2.3.0             | 12.1.1 (EFA)        | 3.11             | `mosaicml/pytorch:latest-aws`, `mosaicml/pytorch:2.3.0_cu121-python3.11-ubuntu20.04-aws` |
-| Ubuntu 20.04   | Base     | 2.3.0             | cpu                 | 3.11             | `mosaicml/pytorch:latest_cpu`, `mosaicml/pytorch:2.3.0_cpu-python3.11-ubuntu20.04`       |
+| Ubuntu 20.04   | Base     | 2.3.1             | 12.1.1 (Infiniband) | 3.11             | `mosaicml/pytorch:latest`, `mosaicml/pytorch:2.3.1_cu121-python3.11-ubuntu20.04`         |
+| Ubuntu 20.04   | Base     | 2.3.1             | 12.1.1 (EFA)        | 3.11             | `mosaicml/pytorch:latest-aws`, `mosaicml/pytorch:2.3.1_cu121-python3.11-ubuntu20.04-aws` |
+| Ubuntu 20.04   | Base     | 2.3.1             | cpu                 | 3.11             | `mosaicml/pytorch:latest_cpu`, `mosaicml/pytorch:2.3.1_cpu-python3.11-ubuntu20.04`       |
 | Ubuntu 20.04   | Base     | 2.2.2             | 12.1.1 (Infiniband) | 3.11             | `mosaicml/pytorch:2.2.2_cu121-python3.11-ubuntu20.04`                                    |
 | Ubuntu 20.04   | Base     | 2.2.2             | 12.1.1 (EFA)        | 3.11             | `mosaicml/pytorch:2.2.2_cu121-python3.11-ubuntu20.04-aws`                                |
 | Ubuntu 20.04   | Base     | 2.2.2             | cpu                 | 3.11             | `mosaicml/pytorch:2.2.2_cpu-python3.11-ubuntu20.04`                                      |

diff --git a/docker/build_matrix.yaml b/docker/build_matrix.yaml
@@ -2,7 +2,7 @@
 - AWS_OFI_NCCL_VERSION: ''
   BASE_IMAGE: nvidia/cuda:12.1.1-cudnn8-devel-ubuntu20.04
   CUDA_VERSION: 12.1.1
-  IMAGE_NAME: torch-2-3-0-cu121
+  IMAGE_NAME: torch-2-3-1-cu121
   MOFED_VERSION: latest-23.10
   NVIDIA_REQUIRE_CUDA_OVERRIDE: cuda>=12.1 brand=tesla,driver>=450,driver<451 brand=tesla,driver>=470,driver<471
     brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471
@@ -21,16 +21,16 @@
   PYTHON_VERSION: '3.11'
   PYTORCH_NIGHTLY_URL: ''
   PYTORCH_NIGHTLY_VERSION: ''
-  PYTORCH_VERSION: 2.3.0
+  PYTORCH_VERSION: 2.3.1
   TAGS:
-  - mosaicml/pytorch:2.3.0_cu121-python3.11-ubuntu20.04
+  - mosaicml/pytorch:2.3.1_cu121-python3.11-ubuntu20.04
   - mosaicml/pytorch:latest
   TARGET: pytorch_stage
-  TORCHVISION_VERSION: 0.18.0
+  TORCHVISION_VERSION: 0.18.1
 - AWS_OFI_NCCL_VERSION: v1.9.1-aws
   BASE_IMAGE: nvidia/cuda:12.1.1-cudnn8-devel-ubuntu20.04
   CUDA_VERSION: 12.1.1
-  IMAGE_NAME: torch-2-3-0-cu121-aws
+  IMAGE_NAME: torch-2-3-1-cu121-aws
   MOFED_VERSION: ''
   NVIDIA_REQUIRE_CUDA_OVERRIDE: cuda>=12.1 brand=tesla,driver>=450,driver<451 brand=tesla,driver>=470,driver<471
     brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471
@@ -49,27 +49,27 @@
   PYTHON_VERSION: '3.11'
   PYTORCH_NIGHTLY_URL: ''
   PYTORCH_NIGHTLY_VERSION: ''
-  PYTORCH_VERSION: 2.3.0
+  PYTORCH_VERSION: 2.3.1
   TAGS:
-  - mosaicml/pytorch:2.3.0_cu121-python3.11-ubuntu20.04-aws
+  - mosaicml/pytorch:2.3.1_cu121-python3.11-ubuntu20.04-aws
   - mosaicml/pytorch:latest-aws
   TARGET: pytorch_stage
-  TORCHVISION_VERSION: 0.18.0
+  TORCHVISION_VERSION: 0.18.1
 - AWS_OFI_NCCL_VERSION: ''
   BASE_IMAGE: ubuntu:20.04
   CUDA_VERSION: ''
-  IMAGE_NAME: torch-2-3-0-cpu
+  IMAGE_NAME: torch-2-3-1-cpu
   MOFED_VERSION: ''
   NVIDIA_REQUIRE_CUDA_OVERRIDE: ''
   PYTHON_VERSION: '3.11'
   PYTORCH_NIGHTLY_URL: ''
   PYTORCH_NIGHTLY_VERSION: ''
-  PYTORCH_VERSION: 2.3.0
+  PYTORCH_VERSION: 2.3.1
   TAGS:
-  - mosaicml/pytorch:2.3.0_cpu-python3.11-ubuntu20.04
+  - mosaicml/pytorch:2.3.1_cpu-python3.11-ubuntu20.04
   - mosaicml/pytorch:latest_cpu
   TARGET: pytorch_stage
-  TORCHVISION_VERSION: 0.18.0
+  TORCHVISION_VERSION: 0.18.1
 - AWS_OFI_NCCL_VERSION: ''
   BASE_IMAGE: nvidia/cuda:12.1.1-cudnn8-devel-ubuntu20.04
   CUDA_VERSION: 12.1.1
@@ -229,12 +229,12 @@
   PYTHON_VERSION: '3.11'
   PYTORCH_NIGHTLY_URL: ''
   PYTORCH_NIGHTLY_VERSION: ''
-  PYTORCH_VERSION: 2.3.0
+  PYTORCH_VERSION: 2.3.1
   TAGS:
   - mosaicml/composer:0.23.0
   - mosaicml/composer:latest
   TARGET: composer_stage
-  TORCHVISION_VERSION: 0.18.0
+  TORCHVISION_VERSION: 0.18.1
 - AWS_OFI_NCCL_VERSION: ''
   BASE_IMAGE: ubuntu:20.04
   COMPOSER_INSTALL_COMMAND: mosaicml[all]==0.23.0
@@ -245,9 +245,9 @@
   PYTHON_VERSION: '3.11'
   PYTORCH_NIGHTLY_URL: ''
   PYTORCH_NIGHTLY_VERSION: ''
-  PYTORCH_VERSION: 2.3.0
+  PYTORCH_VERSION: 2.3.1
   TAGS:
   - mosaicml/composer:0.23.0_cpu
   - mosaicml/composer:latest_cpu
   TARGET: composer_stage
-  TORCHVISION_VERSION: 0.18.0
+  TORCHVISION_VERSION: 0.18.1
diff --git a/docker/generate_build_matrix.py b/docker/generate_build_matrix.py
@@ -19,12 +19,12 @@
 import yaml
 
 PRODUCTION_PYTHON_VERSION = '3.11'
-PRODUCTION_PYTORCH_VERSION = '2.3.0'
+PRODUCTION_PYTORCH_VERSION = '2.3.1'
 
 
 def _get_torchvision_version(pytorch_version: str):
-    if pytorch_version == '2.3.0':
-        return '0.18.0'
+    if pytorch_version == '2.3.1':
+        return '0.18.1'
     if pytorch_version == '2.2.2':
         return '0.17.2'
     if pytorch_version == '2.1.2':
@@ -42,7 +42,7 @@ def _get_cuda_version(pytorch_version: str, use_cuda: bool):
     # From https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/
     if not use_cuda:
         return ''
-    if pytorch_version == '2.3.0':
+    if pytorch_version == '2.3.1':
         return '12.1.1'
     if pytorch_version == '2.2.2':
         return '12.1.1'
@@ -167,7 +167,7 @@ def _write_table(table_tag: str, table_contents: str):
 
 
 def _main():
-    python_pytorch_versions = [('3.11', '2.3.0'), ('3.11', '2.2.2'), ('3.10', '2.1.2')]
+    python_pytorch_versions = [('3.11', '2.3.1'), ('3.11', '2.2.2'), ('3.10', '2.1.2')]
     cuda_options = [True, False]
     stages = ['pytorch_stage']
     interconnects = ['mellanox', 'EFA']  # mellanox is default, EFA needed for AWS