experiment with nvcr pytorch image for torch 2.5.1

.github/workflows/base.yml

@@ -28,30 +28,32 @@ jobs:
             python_version: "3.10"
             pytorch: 2.3.1
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+            from_base_img: ""
+            from_base_tag: ""
           - cuda: "121"
             cuda_version: 12.1.1
             cudnn_version: 8
             python_version: "3.11"
             pytorch: 2.3.1
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
-          - cuda: "124"
-            cuda_version: 12.4.1
-            cudnn_version: ""
-            python_version: "3.10"
-            pytorch: 2.4.1
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+            from_base_img: ""
+            from_base_tag: ""
           - cuda: "124"
             cuda_version: 12.4.1
             cudnn_version: ""
             python_version: "3.11"
             pytorch: 2.4.1
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+            from_base_img: ""
+            from_base_tag: ""
           - cuda: "124"
             cuda_version: 12.4.1
             cudnn_version: ""
             python_version: "3.11"
             pytorch: 2.5.1
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+            from_base_img: nvcr.io/nvidia/pytorch
+            from_base_tag: 24.10-py3
     steps:
       - name: Checkout
         uses: actions/checkout@v4
@@ -84,3 +86,5 @@ jobs:
             PYTHON_VERSION=${{ matrix.python_version }}
             PYTORCH_VERSION=${{ matrix.pytorch }}
             TORCH_CUDA_ARCH_LIST=${{ matrix.torch_cuda_arch_list }}
+            BASE_IMAGE=${{ matrix.from_base_img || '' }}
+            BASE_TAG=${{ matrix.from_base_tag || '' }}

docker/Dockerfile

@@ -1,5 +1,6 @@
+ARG BASE_IMAGE=axolotlai/axolotl-base
 ARG BASE_TAG=main-base
-FROM axolotlai/axolotl-base:$BASE_TAG
+FROM $BASE_IMAGE:$BASE_TAG
 
 ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
 ARG AXOLOTL_EXTRAS=""

docker/Dockerfile-base

@@ -3,7 +3,9 @@ ARG CUDNN_VERSION="8"
 ARG UBUNTU_VERSION="22.04"
 ARG MAX_JOBS=4
 
-FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION AS base-builder
+ARG BASE_IMAGE=nvidia/cuda
+ARG BASE_TAG=""
+FROM $BASE_IMAGE:${BASE_TAG:-$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION} AS base-builder
 
 ENV PATH="/root/miniconda3/bin:${PATH}"