Add docker images for torch 2.4 (#1704)

.github/scripts/docker/generate_build_matrix.py

@@ -45,7 +45,7 @@ def get_torchaudio_version(torch_version):
 def get_matrix():
     k2_version = "1.24.4.dev20240223"
     kaldifeat_version = "1.25.4.dev20240223"
-    version = "20240606"
+    version = "20240725"
     python_version = ["3.8", "3.9", "3.10", "3.11", "3.12"]
     torch_version = []
     torch_version += ["1.13.0", "1.13.1"]
@@ -53,6 +53,7 @@ def get_matrix():
     torch_version += ["2.1.0", "2.1.1", "2.1.2"]
     torch_version += ["2.2.0", "2.2.1", "2.2.2"]
     torch_version += ["2.3.0", "2.3.1"]
+    torch_version += ["2.4.0"]
 
     matrix = []
     for p in python_version:
@@ -78,6 +79,9 @@ def get_matrix():
             elif t == "2.3.1":
                 k2_version_2 = "1.24.4.dev20240606"
                 kaldifeat_version_2 = "1.25.4.dev20240606"
+            elif t == "2.4.0":
+                k2_version_2 = "1.24.4.dev20240725"
+                kaldifeat_version_2 = "1.25.4.dev20240725"
 
             matrix.append(
                 {

.github/workflows/build-docker-image.yml

@@ -16,7 +16,7 @@ jobs:
       fail-fast: false
       matrix:
         os: [ubuntu-latest]
-        image: ["torch2.3.1-cuda12.1", "torch2.3.1-cuda11.8", "torch2.2.2-cuda12.1", "torch2.2.2-cuda11.8", "torch2.2.1-cuda12.1", "torch2.2.1-cuda11.8", "torch2.2.0-cuda12.1", "torch2.2.0-cuda11.8", "torch2.1.0-cuda12.1", "torch2.1.0-cuda11.8", "torch2.0.0-cuda11.7", "torch1.13.0-cuda11.6", "torch1.12.1-cuda11.3", "torch1.9.0-cuda10.2"]
+        image: ["torch2.4.0-cuda12.4", "torch2.4.0-cuda12.1", "torch2.4.0-cuda11.8", "torch2.3.1-cuda12.1", "torch2.3.1-cuda11.8", "torch2.2.2-cuda12.1", "torch2.2.2-cuda11.8", "torch2.2.1-cuda12.1", "torch2.2.1-cuda11.8", "torch2.2.0-cuda12.1", "torch2.2.0-cuda11.8", "torch2.1.0-cuda12.1", "torch2.1.0-cuda11.8", "torch2.0.0-cuda11.7", "torch1.13.0-cuda11.6", "torch1.12.1-cuda11.3", "torch1.9.0-cuda10.2"]
 
     steps:
       # refer to https://github.com/actions/checkout
@@ -37,6 +37,38 @@ jobs:
           rm -rf /opt/hostedtoolcache
           df -h
 
+      - name: Free more space
+        shell: bash
+        run: |
+          # https://github.com/orgs/community/discussions/25678
+          cd /opt
+          find . -maxdepth 1 -mindepth 1 '!' -path ./containerd '!' -path ./actionarchivecache '!' -path ./runner '!' -path ./runner-cache -exec rm -rf '{}' ';'
+
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf "/usr/local/share/boost"
+          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
+
+      - name: Free Disk Space (Ubuntu)
+        uses: jlumbroso/free-disk-space@main
+        with:
+          # this might remove tools that are actually needed,
+          # if set to "true" but frees about 6 GB
+          tool-cache: false
+
+          # all of these default to true, but feel free to set to
+          # "false" if necessary for your workflow
+          android: true
+          dotnet: true
+          haskell: true
+          large-packages: true
+          docker-images: false
+          swap-storage: true
+
+      - name: Check space
+        shell: bash
+        run: |
+          df -h
+
       - name: Log in to Docker Hub
         uses: docker/login-action@v2
         with:

.github/workflows/run-docker-image.yml

@@ -14,7 +14,7 @@ jobs:
       fail-fast: false
       matrix:
         os: [ubuntu-latest]
-        image: ["torch2.3.1-cuda12.1", "torch2.3.1-cuda11.8", "torch2.2.2-cuda12.1", "torch2.2.2-cuda11.8", "torch2.2.1-cuda12.1", "torch2.2.1-cuda11.8", "torch2.2.0-cuda12.1", "torch2.2.0-cuda11.8", "torch2.1.0-cuda12.1", "torch2.1.0-cuda11.8", "torch2.0.0-cuda11.7", "torch1.13.0-cuda11.6", "torch1.12.1-cuda11.3", "torch1.9.0-cuda10.2"]
+        image: ["torch2.4.0-cuda12.4", "torch2.4.0-cuda12.1", "torch2.4.0-cuda11.8", "torch2.3.1-cuda12.1", "torch2.3.1-cuda11.8", "torch2.2.2-cuda12.1", "torch2.2.2-cuda11.8", "torch2.2.1-cuda12.1", "torch2.2.1-cuda11.8", "torch2.2.0-cuda12.1", "torch2.2.0-cuda11.8", "torch2.1.0-cuda12.1", "torch2.1.0-cuda11.8", "torch2.0.0-cuda11.7", "torch1.13.0-cuda11.6", "torch1.12.1-cuda11.3", "torch1.9.0-cuda10.2"]
     steps:
       # refer to https://github.com/actions/checkout
       - uses: actions/checkout@v2
@@ -28,6 +28,38 @@ jobs:
           rm -rf /opt/hostedtoolcache
           df -h
 
+      - name: Free more space
+        shell: bash
+        run: |
+          # https://github.com/orgs/community/discussions/25678
+          cd /opt
+          find . -maxdepth 1 -mindepth 1 '!' -path ./containerd '!' -path ./actionarchivecache '!' -path ./runner '!' -path ./runner-cache -exec rm -rf '{}' ';'
+
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf "/usr/local/share/boost"
+          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
+
+      - name: Free Disk Space (Ubuntu)
+        uses: jlumbroso/free-disk-space@main
+        with:
+          # this might remove tools that are actually needed,
+          # if set to "true" but frees about 6 GB
+          tool-cache: false
+
+          # all of these default to true, but feel free to set to
+          # "false" if necessary for your workflow
+          android: true
+          dotnet: true
+          haskell: true
+          large-packages: true
+          docker-images: false
+          swap-storage: true
+
+      - name: Check space
+        shell: bash
+        run: |
+          df -h
+
       - name: Run the build process with Docker
         uses: addnab/docker-run-action@v3
         with:

docker/torch2.0.0-cuda11.7.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.1.0-cuda11.8.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.1.0-cuda12.1.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.2.0-cuda11.8.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.2.0-cuda12.1.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.2.1-cuda11.8.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.2.1-cuda12.1.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.2.2-cuda11.8.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.2.2-cuda12.1.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.3.1-cuda11.8.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.3.1-cuda12.1.dockerfile

@@ -41,7 +41,7 @@ RUN apt-get update && \
 
 # Install dependencies
 RUN pip install --no-cache-dir \
-      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torch_stable.html \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
       k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
       git+https://github.com/lhotse-speech/lhotse \
       kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \

docker/torch2.4.0-cuda11.8.dockerfile

@@ -0,0 +1,73 @@
+FROM pytorch/pytorch:2.4.0-cuda11.8-cudnn9-devel
+# python 3.10
+
+ENV LC_ALL C.UTF-8
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+# python 3.10
+ARG K2_VERSION="1.24.4.dev20240725+cuda11.8.torch2.4.0"
+ARG KALDIFEAT_VERSION="1.25.4.dev20240725+cuda11.8.torch2.4.0"
+ARG TORCHAUDIO_VERSION="2.4.0+cu118"
+
+LABEL authors="Fangjun Kuang <csukuangfj@gmail.com>"
+LABEL k2_version=${K2_VERSION}
+LABEL kaldifeat_version=${KALDIFEAT_VERSION}
+LABEL github_repo="https://github.com/k2-fsa/icefall"
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+        curl \
+        vim \
+    	libssl-dev \
+        autoconf \
+        automake \
+        bzip2 \
+        ca-certificates \
+        ffmpeg \
+        g++ \
+        gfortran \
+        git \
+        libtool \
+        make \
+        patch \
+        sox \
+        subversion \
+        unzip \
+        valgrind \
+        wget \
+        zlib1g-dev \
+        && rm -rf /var/lib/apt/lists/*
+
+# Install dependencies
+RUN pip install --no-cache-dir \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
+      k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
+      git+https://github.com/lhotse-speech/lhotse \
+      kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
+      kaldi_native_io \
+      kaldialign \
+      kaldifst \
+      kaldilm \
+      sentencepiece>=0.1.96 \
+      tensorboard \
+      typeguard \
+      dill \
+      onnx \
+      onnxruntime \
+      onnxmltools \
+      onnxoptimizer \
+      onnxsim \
+      multi_quantization \
+      typeguard \
+      numpy \
+      pytest \
+      graphviz
+
+RUN git clone https://github.com/k2-fsa/icefall /workspace/icefall && \
+    cd /workspace/icefall && \
+    pip install --no-cache-dir -r requirements.txt
+
+ENV PYTHONPATH /workspace/icefall:$PYTHONPATH
+
+WORKDIR /workspace/icefall

docker/torch2.4.0-cuda12.1.dockerfile

@@ -0,0 +1,73 @@
+FROM pytorch/pytorch:2.4.0-cuda12.1-cudnn9-devel
+# python 3.10
+
+ENV LC_ALL C.UTF-8
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+# python 3.10
+ARG K2_VERSION="1.24.4.dev20240725+cuda12.1.torch2.4.0"
+ARG KALDIFEAT_VERSION="1.25.4.dev20240725+cuda12.1.torch2.4.0"
+ARG TORCHAUDIO_VERSION="2.4.0+cu121"
+
+LABEL authors="Fangjun Kuang <csukuangfj@gmail.com>"
+LABEL k2_version=${K2_VERSION}
+LABEL kaldifeat_version=${KALDIFEAT_VERSION}
+LABEL github_repo="https://github.com/k2-fsa/icefall"
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+        curl \
+        vim \
+    	libssl-dev \
+        autoconf \
+        automake \
+        bzip2 \
+        ca-certificates \
+        ffmpeg \
+        g++ \
+        gfortran \
+        git \
+        libtool \
+        make \
+        patch \
+        sox \
+        subversion \
+        unzip \
+        valgrind \
+        wget \
+        zlib1g-dev \
+        && rm -rf /var/lib/apt/lists/*
+
+# Install dependencies
+RUN pip install --no-cache-dir \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
+      k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
+      git+https://github.com/lhotse-speech/lhotse \
+      kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
+      kaldi_native_io \
+      kaldialign \
+      kaldifst \
+      kaldilm \
+      sentencepiece>=0.1.96 \
+      tensorboard \
+      typeguard \
+      dill \
+      onnx \
+      onnxruntime \
+      onnxmltools \
+      onnxoptimizer \
+      onnxsim \
+      multi_quantization \
+      typeguard \
+      numpy \
+      pytest \
+      graphviz
+
+RUN git clone https://github.com/k2-fsa/icefall /workspace/icefall && \
+    cd /workspace/icefall && \
+    pip install --no-cache-dir -r requirements.txt
+
+ENV PYTHONPATH /workspace/icefall:$PYTHONPATH
+
+WORKDIR /workspace/icefall

docker/torch2.4.0-cuda12.4.dockerfile

@@ -0,0 +1,73 @@
+FROM pytorch/pytorch:2.4.0-cuda12.4-cudnn9-devel
+# python 3.10
+
+ENV LC_ALL C.UTF-8
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+# python 3.10
+ARG K2_VERSION="1.24.4.dev20240725+cuda12.4.torch2.4.0"
+ARG KALDIFEAT_VERSION="1.25.4.dev20240725+cuda12.4.torch2.4.0"
+ARG TORCHAUDIO_VERSION="2.4.0+cu124"
+
+LABEL authors="Fangjun Kuang <csukuangfj@gmail.com>"
+LABEL k2_version=${K2_VERSION}
+LABEL kaldifeat_version=${KALDIFEAT_VERSION}
+LABEL github_repo="https://github.com/k2-fsa/icefall"
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+        curl \
+        vim \
+    	libssl-dev \
+        autoconf \
+        automake \
+        bzip2 \
+        ca-certificates \
+        ffmpeg \
+        g++ \
+        gfortran \
+        git \
+        libtool \
+        make \
+        patch \
+        sox \
+        subversion \
+        unzip \
+        valgrind \
+        wget \
+        zlib1g-dev \
+        && rm -rf /var/lib/apt/lists/*
+
+# Install dependencies
+RUN pip install --no-cache-dir \
+      torchaudio==${TORCHAUDIO_VERSION} -f https://download.pytorch.org/whl/torchaudio/ \
+      k2==${K2_VERSION} -f https://k2-fsa.github.io/k2/cuda.html \
+      git+https://github.com/lhotse-speech/lhotse \
+      kaldifeat==${KALDIFEAT_VERSION} -f https://csukuangfj.github.io/kaldifeat/cuda.html \
+      kaldi_native_io \
+      kaldialign \
+      kaldifst \
+      kaldilm \
+      sentencepiece>=0.1.96 \
+      tensorboard \
+      typeguard \
+      dill \
+      onnx \
+      onnxruntime \
+      onnxmltools \
+      onnxoptimizer \
+      onnxsim \
+      multi_quantization \
+      typeguard \
+      numpy \
+      pytest \
+      graphviz
+
+RUN git clone https://github.com/k2-fsa/icefall /workspace/icefall && \
+    cd /workspace/icefall && \
+    pip install --no-cache-dir -r requirements.txt
+
+ENV PYTHONPATH /workspace/icefall:$PYTHONPATH
+
+WORKDIR /workspace/icefall