Merge branch 'master' into move_export

Author: chensuyue

.azure-pipelines/model-test-3x.yml

@@ -14,6 +14,7 @@ pr:
       - setup.py
       - requirements_pt.txt
       - .azure-pipelines/scripts/models
+      - .azure-pipelines/model-test-3x.yml
 
 variables:
   OUT_SCRIPT_PATH: $(Build.SourcesDirectory)/.azure-pipelines/scripts/models
@@ -35,7 +36,7 @@ parameters:
 stages:
   - stage: PyTorchModels
     displayName: Run PyTorch Model
-    pool: MODEL_PERF_TEST
+    pool: ICX-16C
     dependsOn: []
     condition: and(succeeded(), eq('${{ parameters.PyTorch_Model_3X }}', 'true'))
     jobs:
@@ -60,7 +61,7 @@ stages:
           - script: |
               echo ${BUILD_SOURCESDIRECTORY}
               rm -fr ${BUILD_SOURCESDIRECTORY} || sudo rm -fr ${BUILD_SOURCESDIRECTORY} || true
-              echo y | docker system prune
+              echo y | docker system prune --all
             displayName: "Clean workspace"
           - checkout: self
             clean: true

.azure-pipelines/model-test.yml

@@ -13,6 +13,7 @@ pr:
       - requirements.txt
       - .azure-pipelines/scripts/models
       - examples/tensorflow/oob_models/quantization/ptq
+      - .azure-pipelines/model-test.yml
     exclude:
       - test
       - neural_compressor/common
@@ -49,10 +50,10 @@ parameters:
     default:
       - resnet50v1.5
       - ssd_resnet50_v1
-      - ssd_mobilenet_v1_ckpt
-      - inception_v1
-      - darknet19
-      - resnet-101
+#      - ssd_mobilenet_v1_ckpt
+#      - inception_v1
+#      - darknet19
+#      - resnet-101
   - name: PyTorchModelList
     type: object
     default:
@@ -70,7 +71,7 @@ parameters:
 stages:
   - stage: TensorFlowModels
     displayName: Run TensorFlow Model
-    pool: MODEL_PERF_TEST_TF
+    pool: MODEL_PERF_TEST
     dependsOn: []
     condition: and(succeeded(), eq('${{ parameters.TensorFlow_Model }}', 'true'))
     jobs:
@@ -139,7 +140,7 @@ stages:
           - script: |
               echo ${BUILD_SOURCESDIRECTORY}
               rm -fr ${BUILD_SOURCESDIRECTORY} || sudo rm -fr ${BUILD_SOURCESDIRECTORY} || true
-              echo y | docker system prune
+              echo y | docker system prune --all
             displayName: "Clean workspace"
           - checkout: self
             clean: true

.azure-pipelines/scripts/fwk_version.sh

@@ -4,21 +4,21 @@ echo "export FWs version..."
 test_mode=$1
 
 if [ "$test_mode" == "coverage" ] || [ "$test_mode" == "latest" ]; then
+    export tensorflow_version='2.15.0-official'
+    export pytorch_version='2.3.0+cpu'
+    export torchvision_version='0.18.0+cpu'
+    export ipex_version='2.3.0+cpu'
+    export onnx_version='1.16.0'
+    export onnxruntime_version='1.18.0'
+    export mxnet_version='1.9.1'
+else
     export tensorflow_version='2.15.0-official'
     export pytorch_version='2.2.1+cpu'
     export torchvision_version='0.17.1+cpu'
     export ipex_version='2.2.0+cpu'
     export onnx_version='1.15.0'
     export onnxruntime_version='1.17.1'
     export mxnet_version='1.9.1'
-else
-    export tensorflow_version='2.14.0'
-    export pytorch_version='2.1.0+cpu'
-    export torchvision_version='0.16.0+cpu'
-    export ipex_version='2.1.0+cpu'
-    export onnx_version='1.14.1'
-    export onnxruntime_version='1.16.3'
-    export mxnet_version='1.9.1'
 fi
 
 

.azure-pipelines/scripts/ut/run_basic_itex.sh

@@ -4,10 +4,10 @@ test_case="run basic itex"
 echo "${test_case}"
 
 echo "specify fwk version..."
-export itex_version='2.14.0.1'
-export tensorflow_version='2.14.0-official'
-export onnx_version='1.14.1'
-export onnxruntime_version='1.16.3'
+export itex_version='2.15.0.0'
+export tensorflow_version='2.15.0-official'
+export onnx_version='1.16.0'
+export onnxruntime_version='1.18.0'
 
 echo "set up UT env..."
 bash /neural-compressor/.azure-pipelines/scripts/ut/env_setup.sh "${test_case}"

.azure-pipelines/scripts/ut/run_basic_pt_pruning.sh

@@ -4,9 +4,9 @@ test_case="run basic pt pruning"
 echo "${test_case}"
 
 echo "specify fwk version..."
-export pytorch_version='2.1.0+cpu'
-export torchvision_version='0.16.0+cpu'
-export ipex_version='2.1.0+cpu'
+export pytorch_version='2.3.0+cpu'
+export torchvision_version='0.18.0+cpu'
+export ipex_version='2.3.0+cpu'
 
 echo "set up UT env..."
 bash /neural-compressor/.azure-pipelines/scripts/ut/env_setup.sh "${test_case}"

.azure-pipelines/template/docker-template.yml

@@ -33,7 +33,7 @@ steps:
   - ${{ if eq(parameters.dockerConfigName, 'commonDockerConfig') }}:
       - script: |
           rm -fr ${BUILD_SOURCESDIRECTORY} || sudo rm -fr ${BUILD_SOURCESDIRECTORY} || true
-          echo y | docker system prune
+          echo y | docker system prune --all
         displayName: "Clean workspace"
 
       - checkout: self
@@ -45,7 +45,7 @@ steps:
           rm -fr ${BUILD_SOURCESDIRECTORY} || sudo rm -fr ${BUILD_SOURCESDIRECTORY} || true
           mkdir ${BUILD_SOURCESDIRECTORY}
           chmod 777 ${BUILD_SOURCESDIRECTORY}
-          echo y | docker system prune
+          echo y | docker system prune --all
         displayName: "Clean workspace"
 
       - checkout: none

README.md

@@ -195,8 +195,9 @@ quantized_model = fit(model=float_model, conf=static_quant_conf, calib_dataloade
 > Further documentations can be found at [User Guide](https://github.com/intel/neural-compressor/blob/master/docs/source/user_guide.md).
 
 ## Selected Publications/Events
-* Blog by Intel: [Accelerate Meta* Llama 3 with Intel AI Solutions](https://www.intel.com/content/www/us/en/developer/articles/technical/accelerate-meta-llama3-with-intel-ai-solutions.html)
-* Blog by Intel: [Effective Weight-Only Quantization for Large Language Models with Intel® Neural Compressor](https://community.intel.com/t5/Blogs/Tech-Innovation/Artificial-Intelligence-AI/Effective-Weight-Only-Quantization-for-Large-Language-Models/post/1529552) (Oct 2023)
+* Blog by Intel: [Neural Compressor: Boosting AI Model Efficiency](https://community.intel.com/t5/Blogs/Tech-Innovation/Artificial-Intelligence-AI/Neural-Compressor-Boosting-AI-Model-Efficiency/post/1604740) (June 2024) 
+* Blog by Intel: [Optimization of Intel AI Solutions for Alibaba Cloud’s Qwen2 Large Language Models](https://www.intel.com/content/www/us/en/developer/articles/technical/intel-ai-solutions-accelerate-alibaba-qwen2-llms.html) (June 2024)
+* Blog by Intel: [Accelerate Meta* Llama 3 with Intel AI Solutions](https://www.intel.com/content/www/us/en/developer/articles/technical/accelerate-meta-llama3-with-intel-ai-solutions.html) (Apr 2024)
 * EMNLP'2023 (Under Review): [TEQ: Trainable Equivalent Transformation for Quantization of LLMs](https://openreview.net/forum?id=iaI8xEINAf&referrer=%5BAuthor%20Console%5D) (Sep 2023)
 * arXiv: [Efficient Post-training Quantization with FP8 Formats](https://arxiv.org/abs/2309.14592) (Sep 2023)
 * arXiv: [Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs](https://arxiv.org/abs/2309.05516) (Sep 2023)

docker/Dockerfile

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-ARG UBUNTU_VER=20.04
+ARG UBUNTU_VER=22.04
 FROM ubuntu:${UBUNTU_VER} as deploy
 
 # See http://bugs.python.org/issue19846

docs/source/publication_list.md

@@ -1,6 +1,8 @@
-Full Publications/Events (80)
+Full Publications/Events (82)
 ==========
-## 2024 (1)
+## 2024 (3)
+* Blog by Intel: [Neural Compressor: Boosting AI Model Efficiency](https://community.intel.com/t5/Blogs/Tech-Innovation/Artificial-Intelligence-AI/Neural-Compressor-Boosting-AI-Model-Efficiency/post/1604740) (June 2024) 
+* Blog by Intel: [Optimization of Intel AI Solutions for Alibaba Cloud’s Qwen2 Large Language Models](https://www.intel.com/content/www/us/en/developer/articles/technical/intel-ai-solutions-accelerate-alibaba-qwen2-llms.html) (June 2024)
 * Blog by Intel: [Accelerate Meta* Llama 3 with Intel AI Solutions](https://www.intel.com/content/www/us/en/developer/articles/technical/accelerate-meta-llama3-with-intel-ai-solutions.html) (Apr 2024)
 
 ## 2023 (25)

examples/3.x_api/pytorch/nlp/huggingface_models/language-modeling/quantization/llm/run_clm_no_trainer.py

@@ -276,7 +276,7 @@ def get_user_model():
                 max_seq_length=args.gptq_max_seq_length,
             )
             dataloader_for_calibration = dataloaderPreprocessor.get_prepared_dataloader()
-            from neural_compressor.torch.algorithms.weight_only.gptq import move_input_to_device
+            from neural_compressor.torch.algorithms.weight_only.utility import move_input_to_device
             from tqdm import tqdm
             def run_fn_for_gptq(model, dataloader_for_calibration, *args):
                 for batch in tqdm(dataloader_for_calibration):