add requirements

.github/workflows/deepspeed.yaml

@@ -7,13 +7,15 @@ on:
   workflow_dispatch:
   pull_request:
     paths:
-      - '.github/workflows/deepspeed.yml'
+      - '.github/workflows/deepspeed.yaml'
+      - '.github/workflows/deepspeed.yaml'
       - 'requirements/**'
   schedule:
     - cron: "0 0 * * *"
   push:
     paths:
-      - '.github/workflows/deepspeed.yml'
+      - '.github/workflows/deepspeed.yaml'
+      - '.github/workflows/deepspeed.yaml'
 
 
 concurrency:
@@ -39,31 +41,26 @@ jobs:
         - /etc/ascend_install.info:/etc/ascend_install.info   
       options: --network host
                --name deepspeed_unit-tests
-               --device /dev/davinci5
+               --device /dev/davinci6
                --device /dev/davinci_manager
                --device /dev/devmm_svm
                --device /dev/hisi_hdc
                --shm-size "20g"
                --entrypoint /bin/bash
-
-    env:
-      PT_HPU_LAZY_MODE: 0
-      TORCHINDUCTOR_COMPILE_THREADS: 1
-      TEST_LIST: |
-        test_accelerator.py
-        test_autotuning.py
-        test_compression.py
-               
+
+
     steps:
     - uses: actions/checkout@v4
 
     - name: Install pytorch 
       run: |
         npu-smi info
-        apt-get update   
+        apt-get update 
+        apt-get install sudo
+        pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
         source /root/.bashrc
 
-        pip install torch==2.2.0 torchvision==0.17.0 torch_npu==2.2.0 numpy==1.26.4 cloudpickle tornado ml-dtypes -i https://pypi.tuna.tsinghua.edu.cn/simple
+        pip install torch==2.2.0 torchvision==0.17.0 torch_npu==2.2.0 torchaudio==2.2.0 numpy==1.26.4 cloudpickle tornado ml-dtypes
         
         python << EOF 
         if __name__ == '__main__':
@@ -74,24 +71,30 @@ jobs:
             print(f"Device Count: {torch.npu.device_count()}")
             print(f"Device Available: {torch.npu.is_available()}")
         EOF
-        
-    # - name: Install transformers
-    #   run: |
-    #     source /root/.bashrc
-    #     echo "y" | apt-get install git
-    #     git clone https://github.com/huggingface/transformers
-    #     cd transformers
-    #     git rev-parse --short HEAD
-    #     pip install . -i https://pypi.tuna.tsinghua.edu.cn/simple
-
-    - name: Install deepspeed
+
+    - name: Install transformers
       run: |
         source /root/.bashrc
         echo "y" | apt-get install git
-        git clone --depth=1 https://github.com/microsoft/DeepSpeed.git
-        cd DeepSpeed
-        pip install .[dev,autotuning] -i  https://pypi.tuna.tsinghua.edu.cn/simple
-        ds_report
+        git clone https://github.com/huggingface/transformers
+        cd transformers
+        git rev-parse --short HEAD
+        pip install .
+             
+    - name: Install deepspeed
+      uses: nick-fields/retry@v3
+      with:
+        timeout_minutes: 30
+        max_attempts: 3
+        retry_on: error
+        command: | 
+          source /root/.bashrc
+          git clone --depth=1 https://github.com/microsoft/DeepSpeed.git
+          pip install -r requirements/requirements_deepspeed.txt 
+          cd DeepSpeed
+          pip install .[1bit,autotuning,inf]   
+  
+          ds_report
 
     - name: Python environment
       run: |
@@ -102,9 +105,29 @@ jobs:
       run: |
         source /root/.bashrc
         unset TORCH_CUDA_ARCH_LIST # only jit compile for current arch
+        cd DeepSpeed/tests/unit/
         
-        cd /DeepSpeed
-        pytest --verbose tests/*
+        pytest --verbose accelerator/*
+        pytest --verbose autotuning/*
+        pytest --verbose checkpoint/test_reshape_checkpoint.py
+        pytest --verbose checkpoint/test_moe_checkpoint.py
+        pytest --verbose checkpoint/test_shared_weights.py
+        pytest --verbose launcher/test_ds_arguments.py launcher/test_run.py
+        pytest --verbose model_parallelism/*
+        pytest --verbose moe/test_moe_tp.py
+        pytest --verbose monitor/*
+        pytest --verbose utils/*
+        pytest --verbose runtime/test_ds_config_model.py
+        pytest --verbose runtime/pipe/test_pipe_schedule.py
+        pytest --verbose runtime/zero/test_zero_config.py
+        pytest --verbose runtime/zero/test_zero_tiled.py
+        pytest --verbose runtime/zero/test_zeropp.py
+        pytest --verbose runtime/test_autocast.py
+        pytest --verbose runtime/test_data.py
+        pytest --verbose runtime/test_runtime_utils.py
+        pytest --verbose runtime/activation_checkpointing/*
+        pytest --verbose runtime/utils/*
+        pytest --verbose runtime/zero/test_zero_dynamic_class.py
 
 
 

README.md

@@ -44,5 +44,20 @@ This [PR](https://github.com/microsoft/onnxruntime/pull/15833) refactored the Ex
 Update on 2023.06.08  
 This [PR](https://github.com/microsoft/onnxruntime/pull/14731) introduced a missing registration of CANN Identity operator for version greater than 14. It has been fixed in this [PR](https://github.com/microsoft/onnxruntime/pull/16210).
 
+## Deepspeed Ascend CI
+The Deepspeed source code is from `main` branch of `microsoft/deepspeed` and will be run and tested daily with Ascend related.
+
+------------------------------------------------------------
+
+| Key  | Value |
+| :---: | :---: |
+| CPU  | Arrch64 |
+| NPU | Ascend910B |
+| OS | Ubantu |
+| Period | UTC 1200 daily |
+| Branch  | main |
+| Status  | ![Deepspeed](https://github.com/Ascend/Ascend-CI/actions/workflows/deepspeed.yaml/badge.svg) |
+| Recheck By Hand | comment 'recheck' in any issue |
+
 ## Pytorch Ascend CI
 TBD

requirements/requirements_deepspeed.txt

@@ -0,0 +1,29 @@
+accelerate
+clang-format==18.1.3
+comet_ml>=3.41.0
+# deepspeed-kernels ; sys_platform == 'linux'
+docutils<0.18
+future
+importlib-metadata>=4
+mup
+pre-commit>=2.20.0
+pytest>=7.2.0
+pytest-forked
+pytest-randomly
+pytest-xdist
+qtorch==0.3.0
+recommonmark
+sphinx
+sphinx-rtd-theme
+tensorboard
+torchvision
+transformers>=4.39.0
+wandb
+# google
+# lm-eval==0.3.0
+# protobuf
+# qtorch
+# safetensors
+# sentencepiece
+# transformers>=4.32.1
+# tabulate