diff --git a/automation/script/module.py b/automation/script/module.py
index e79b33a3e..d72d28493 100644
--- a/automation/script/module.py
+++ b/automation/script/module.py
@@ -975,7 +975,7 @@ def _run(self, i):
                 if str(state['docker'].get('run', True)).lower() in ['false', '0', 'no']:
                     print (recursion_spaces+'  - Skipping script::{} run as we are inside docker'.format(found_script_artifact))
                     return {'return': 0}
-                elif str(state['docker'].get('docker_real_run', True)).lower() in ['false', '0', 'no']:
+                elif str(state['docker'].get('real_run', True)).lower() in ['false', '0', 'no']:
                     print (recursion_spaces+'  - Doing fake run for script::{} as we are inside docker'.format(found_script_artifact))
                     fake_run = True
                     env['CM_TMP_FAKE_RUN']='yes'
diff --git a/script/add-custom-nvidia-system/_cm.yaml b/script/add-custom-nvidia-system/_cm.yaml
index 6705c3cdc..6dce8414d 100644
--- a/script/add-custom-nvidia-system/_cm.yaml
+++ b/script/add-custom-nvidia-system/_cm.yaml
@@ -6,7 +6,8 @@ automation_alias: script
 automation_uid: 5b4e0237da074764
 
 category: "MLPerf benchmark support"
-
+docker:
+  real_run: False
 
 # User-friendly tags to find this CM script
 tags:
@@ -94,6 +95,13 @@ variations:
     add_deps_recursive:
       nvidia-inference-common-code:
         tags: _ctuning
+  go:
+    group: code
+    add_deps_recursive:
+      nvidia-inference-common-code:
+        tags: _go
+
+
 
 
 versions:
@@ -111,3 +119,8 @@ versions:
     add_deps_recursive:
       nvidia-inference-common-code:
         version: r3.1
+
+  r4.0:
+    add_deps_recursive:
+      nvidia-inference-common-code:
+        version: r4.0
diff --git a/script/app-mlperf-inference-ctuning-cpp-tflite/_cm.json b/script/app-mlperf-inference-ctuning-cpp-tflite/_cm.json
index 17caa8047..1d00f3c81 100644
--- a/script/app-mlperf-inference-ctuning-cpp-tflite/_cm.json
+++ b/script/app-mlperf-inference-ctuning-cpp-tflite/_cm.json
@@ -118,6 +118,10 @@
       "tags": "get,ml-model,efficientnet,raw,_tflite"
     },
     {
+      "names": [
+        "tensorflow",
+	"tflite"
+      ],
       "tags": "get,tensorflow,lib,_tflite"
     },
     {
diff --git a/script/app-mlperf-inference-mlcommons-python/customize.py b/script/app-mlperf-inference-mlcommons-python/customize.py
index ee97038ff..27b03b27a 100644
--- a/script/app-mlperf-inference-mlcommons-python/customize.py
+++ b/script/app-mlperf-inference-mlcommons-python/customize.py
@@ -177,7 +177,7 @@ def get_run_cmd_reference(os_info, env, scenario_extra_options, mode_extra_optio
 
         env['RUN_DIR'] = os.path.join(env['CM_MLPERF_INFERENCE_SOURCE'], "language", "gpt-j")
         cmd = env['CM_PYTHON_BIN_WITH_PATH'] +  \
-            " run.py --model-path=" + env['CM_ML_MODEL_FILE_WITH_PATH'] + ' --dataset-path=' + env['CM_DATASET_EVAL_PATH'] + " --scenario " + env['CM_MLPERF_LOADGEN_SCENARIO'] + " " + env['CM_MLPERF_LOADGEN_EXTRA_OPTIONS'] + \
+            " main.py --model-path=" + env['CM_ML_MODEL_FILE_WITH_PATH'] + ' --dataset-path=' + env['CM_DATASET_EVAL_PATH'] + " --scenario " + env['CM_MLPERF_LOADGEN_SCENARIO'] + " " + env['CM_MLPERF_LOADGEN_EXTRA_OPTIONS'] + \
             ' --dtype ' + env['CM_MLPERF_MODEL_PRECISION'] + \
             scenario_extra_options + mode_extra_options + dataset_options
         cmd = cmd.replace("--count", "--max_examples")
diff --git a/script/app-mlperf-inference-nvidia/_cm.yaml b/script/app-mlperf-inference-nvidia/_cm.yaml
index 1dd39f879..86abccbeb 100644
--- a/script/app-mlperf-inference-nvidia/_cm.yaml
+++ b/script/app-mlperf-inference-nvidia/_cm.yaml
@@ -261,7 +261,10 @@ deps:
       CM_MLPERF_NVIDIA_HARNESS_RUN_MODE:
       - run_harness
 
-  - tags: get,generic-python-lib,_package.nvmitten,_path./opt/nvmitten-0.1.3-cp38-cp38-linux_x86_64.whl
+  - tags: get,generic-python-lib,_package.nvmitten
+    update_tags_from_env_with_prefix:
+      _path.:
+        - CM_ENV_NVMITTEN_DOCKER_WHEEL_PATH
     enable_if_env:
       CM_RUN_STATE_DOCKER:
         - 'yes'
@@ -308,6 +311,23 @@ post_deps:
 
 # Variations to customize dependencies
 variations:
+  # MLPerf inference version
+  v4.0:
+    group: version
+    default: true
+    env:
+      CM_MLPERF_INFERENCE_VERSION: "v4.0"
+      CM_MLPERF_GPTJ_MODEL_FP8_PATH_SUFFIX: GPTJ-FP8-quantized
+    adr:
+      pytorch:
+        tags: _for-nvidia-mlperf-inference-v4.0
+  v3.1:
+    env:
+      CM_MLPERF_INFERENCE_VERSION: "v3.1"
+      CM_MLPERF_GPTJ_MODEL_FP8_PATH_SUFFIX: GPTJ-07142023.pth
+    adr:
+      pytorch:
+        tags: _for-nvidia-mlperf-inference-v3.1
   # Target devices
   cpu:
     group: device
@@ -366,6 +386,7 @@ variations:
     - tags: get,generic-python-lib,_transformers
     - tags: get,generic-python-lib,_safetensors
     - tags: get,generic-python-lib,_onnx
+    - tags: get,generic-python-lib,_onnx-graphsurgeon
 
   bert-99:
     group: model
@@ -476,18 +497,25 @@ variations:
     deps:
     - tags: get,generic-python-lib,_package.datasets
     - tags: get,generic-python-lib,_package.simplejson
+    - tags: get,generic-python-lib,_onnx
+    - tags: get,generic-python-lib,_transformers
+    - tags: get,generic-python-lib,_onnx-graphsurgeon
     env:
       CM_ML_MODEL_STARTING_WEIGHTS_FILENAME: "https://cloud.mlcommons.org/index.php/s/QAZ2oM94MkFtbQx/download"
   
   gptj_,build:
     deps:
-    - tags: install,pytorch,from.src,_for-nvidia-mlperf-inference-v3.1
+    - tags: install,pytorch,from.src
+      names:
+      - pytorch
     - tags: get,cmake
       version_min: "3.25.0"
 
   gptj_,build_engine:
     deps:
-    - tags: install,pytorch,from.src,_for-nvidia-mlperf-inference-v3.1
+    - tags: install,pytorch,from.src
+      names:
+      - pytorch
     - tags: get,cmake
       version_min: "3.25.0"
 
@@ -877,7 +905,9 @@ variations:
 
   gptj_,run_harness:
     deps:
-    - tags: install,pytorch,from.src,_for-nvidia-mlperf-inference-v3.1
+    - tags: install,pytorch,from.src
+      names:
+      - pytorch
     - tags: get,cmake
       version_min: "3.25.0"
     env:
diff --git a/script/app-mlperf-inference-nvidia/customize.py b/script/app-mlperf-inference-nvidia/customize.py
index 2d995af9c..bbb9cb703 100644
--- a/script/app-mlperf-inference-nvidia/customize.py
+++ b/script/app-mlperf-inference-nvidia/customize.py
@@ -10,6 +10,9 @@ def preprocess(i):
         return {'return':1, 'error': 'Windows is not supported in this script yet'}
     env = i['env']
 
+    if str(env.get('CM_RUN_STATE_DOCKER', '')).lower() in ['1', 'true', 'yes']:
+        return {'return': 0}
+
     if env.get('CM_MODEL', '') == '':
         return {'return': 1, 'error': 'Please select a variation specifying the model to run'}
 
@@ -153,7 +156,7 @@ def preprocess(i):
             cmds.append("make download_data BENCHMARKS='gptj'")
 
         fp32_model_path = os.path.join(env['MLPERF_SCRATCH_PATH'], 'models', 'GPTJ-6B', 'checkpoint-final')
-        fp8_model_path = os.path.join(env['MLPERF_SCRATCH_PATH'], 'models', 'GPTJ-6B', 'fp8-quantized-ammo', 'GPTJ-07142023.pth')
+        fp8_model_path = os.path.join(env['MLPERF_SCRATCH_PATH'], 'models', 'GPTJ-6B', 'fp8-quantized-ammo', env['CM_MLPERF_GPTJ_MODEL_FP8_PATH_SUFFIX'])
         vocab_path = os.path.join(env['MLPERF_SCRATCH_PATH'], 'models', 'bert', 'vocab.txt')
 
         if not os.path.exists(os.path.dirname(fp32_model_path)):
@@ -163,7 +166,8 @@ def preprocess(i):
 
         if not os.path.exists(fp32_model_path):
             env['CM_REQUIRE_GPTJ_MODEL_DOWNLOAD'] = 'yes' # download via prehook_deps
-            cmds.append(f"cp -r $CM_ML_MODEL_FILE_WITH_PATH {fp32_model_path}")
+            if make_command == "build_engine":
+                cmds.append(f"cp -r $CM_ML_MODEL_FILE_WITH_PATH {fp32_model_path}")
 
         model_name = "gptj"
         model_path = fp8_model_path
diff --git a/script/app-mlperf-inference/_cm.yaml b/script/app-mlperf-inference/_cm.yaml
index c8ff71c11..58f1ca6cb 100644
--- a/script/app-mlperf-inference/_cm.yaml
+++ b/script/app-mlperf-inference/_cm.yaml
@@ -182,6 +182,8 @@ variations:
         tags: _float32
       librispeech-accuracy-script:
         tags: _int32
+      cnndm-accuracy-script:
+        tags: _int32
     env:
       CM_MLPERF_PYTHON: 'yes'
       CM_MLPERF_IMPLEMENTATION: mlcommons_python
@@ -189,6 +191,7 @@ variations:
       CM_IMAGENET_ACCURACY_DTYPE: float32
       CM_OPENIMAGES_ACCURACY_DTYPE: float32
       CM_LIBRISPEECH_ACCURACY_DTYPE: float32
+      CM_CNNDM_ACCURACY_DTYPE: int32
     prehook_deps:
       - names:
          - python-reference-mlperf-inference
@@ -235,6 +238,10 @@ variations:
     default_variations:
       backend: onnxruntime
 
+  nvidia-original,r4.1_default:
+    docker:
+      base_image: nvcr.io/nvidia/mlperf/mlperf-inference:mlpinf-v4.0-cuda12.2-cudnn8.9-x86_64-ubuntu20.04-public
+
   nvidia-original:
     docker:
       interactive: True
@@ -273,12 +280,6 @@ variations:
       CM_IMAGENET_ACCURACY_DTYPE: int32
       CM_CNNDM_ACCURACY_DTYPE: int32
       CM_LIBRISPEECH_ACCURACY_DTYPE: int8
-    deps:
-      - tags: get,cuda-devices
-        skip_if_env:
-          CM_CUDA_DEVICE_PROP_GLOBAL_MEMORY:
-            - "yes"
-            - "on"
     prehook_deps:
       - names:
          - nvidia-original-mlperf-inference
@@ -904,6 +905,12 @@ variations:
     add_deps_recursive:
       mlperf-inference-implementation:
         tags: _cuda
+    deps:
+      - tags: get,cuda-devices
+        skip_if_env:
+          CM_CUDA_DEVICE_PROP_GLOBAL_MEMORY:
+            - "yes"
+            - "on"
   rocm:
     docker:
       all_gpus: 'yes'
@@ -1142,6 +1149,25 @@ variations:
     default_env:
       CM_SKIP_SYS_UTILS: 'yes'
       CM_REGENERATE_MEASURE_FILES: 'yes'
+    env:
+      CM_ENV_NVMITTEN_DOCKER_WHEEL_PATH: '/opt/nvmitten-0.1.3-cp38-cp38-linux_x86_64.whl'
+
+  r4.1_default:
+    group:
+      reproducibility
+    add_deps_recursive:
+      nvidia-inference-common-code:
+        version: r4.0
+        tags: _go
+      nvidia-inference-server:
+        version: r4.0
+        tags: _go
+    default_env:
+      CM_SKIP_SYS_UTILS: 'yes'
+      CM_REGENERATE_MEASURE_FILES: 'yes'
+    env:
+      CM_ENV_NVMITTEN_DOCKER_WHEEL_PATH: '/opt/nvmitten-0.1.3b0-cp38-cp38-linux_x86_64.whl'
+
 
 invalid_variation_combinations:
   -
@@ -1240,6 +1266,7 @@ docker:
     - tags: get,mlperf,inference,results,dir
     - tags: get,mlperf,inference,submission,dir
   pre_run_cmds:
+    #- cm pull repo && cm run script --tags=get,git,repo,_repo.https://github.com/GATEOverflow/inference_results_v4.0.git --update
     - cm pull repo
   mounts:
    - "${{ CM_DATASET_IMAGENET_PATH }}:${{ CM_DATASET_IMAGENET_PATH }}"
diff --git a/script/authenticate-github-cli/_cm.yaml b/script/authenticate-github-cli/_cm.yaml
new file mode 100644
index 000000000..605cc955f
--- /dev/null
+++ b/script/authenticate-github-cli/_cm.yaml
@@ -0,0 +1,16 @@
+alias: authenticate-github-cli
+automation_alias: script
+automation_uid: 5b4e0237da074764
+cache: true
+input_mapping:
+  with_token: CM_GH_AUTH_TOKEN
+  with-token: CM_GH_AUTH_TOKEN
+tags:
+- auth
+- authenticate
+- github
+- gh
+- cli
+uid: 7b57673ac14a4337
+deps:
+  - tags: get,gh,cli
diff --git a/script/authenticate-github-cli/customize.py b/script/authenticate-github-cli/customize.py
new file mode 100644
index 000000000..14b762e0f
--- /dev/null
+++ b/script/authenticate-github-cli/customize.py
@@ -0,0 +1,27 @@
+from cmind import utils
+import os
+
+def preprocess(i):
+
+    os_info = i['os_info']
+
+    env = i['env']
+
+    meta = i['meta']
+
+    automation = i['automation']
+
+    cmd = "gh auth login"
+    if env.get('CM_GH_AUTH_TOKEN', '') != '':
+        cmd = f" echo {env['CM_GH_AUTH_TOKEN']} | {cmd} --with-token"
+
+    env['CM_RUN_CMD'] = cmd
+    quiet = (env.get('CM_QUIET', False) == 'yes')
+
+    return {'return':0}
+
+def postprocess(i):
+
+    env = i['env']
+
+    return {'return':0}
diff --git a/script/authenticate-github-cli/run.bat b/script/authenticate-github-cli/run.bat
new file mode 100644
index 000000000..648302ca7
--- /dev/null
+++ b/script/authenticate-github-cli/run.bat
@@ -0,0 +1 @@
+rem native script
diff --git a/script/authenticate-github-cli/run.sh b/script/authenticate-github-cli/run.sh
new file mode 100644
index 000000000..b17d52e4b
--- /dev/null
+++ b/script/authenticate-github-cli/run.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+
+#CM Script location: ${CM_TMP_CURRENT_SCRIPT_PATH}
+
+#To export any variable
+#echo "VARIABLE_NAME=VARIABLE_VALUE" >>tmp-run-env.out
+
+#${CM_PYTHON_BIN_WITH_PATH} contains the path to python binary if "get,python" is added as a dependency
+
+echo "Running: "
+echo "${CM_RUN_CMD}"
+echo ""
+
+if [[ ${CM_FAKE_RUN} != "yes" ]]; then
+  eval "${CM_RUN_CMD}"
+  test $? -eq 0 || exit 1
+fi
+
diff --git a/script/build-mlperf-inference-server-nvidia/_cm.yaml b/script/build-mlperf-inference-server-nvidia/_cm.yaml
index 98ee5b922..dc2e2fc5b 100644
--- a/script/build-mlperf-inference-server-nvidia/_cm.yaml
+++ b/script/build-mlperf-inference-server-nvidia/_cm.yaml
@@ -164,6 +164,11 @@ variations:
     add_deps_recursive:
       nvidia-inference-common-code:
         tags: _ctuning
+  go:
+    group: code
+    add_deps_recursive:
+      nvidia-inference-common-code:
+        tags: _go
   nvidia-only:
     group: code
     add_deps_recursive:
@@ -180,6 +185,22 @@ variations:
       nvidia-inference-common-code:
         tags: _mlcommons
 
+  r4.0:
+    group: version
+    add_deps_recursive:
+      nvidia-inference-common-code:
+        version: r4.0
+      nvidia-scratch-space:
+        tags: _version.4_1
+    deps:
+      - tags: install,pytorch,from.src,_for-nvidia-mlperf-inference-v4.0
+        names:
+        - pytorch
+        - torch
+      - tags: install,torchvision,from.src,_for-nvidia-mlperf-inference-v4.0
+        names:
+        - pytorchvision
+        - torchvision
 
 versions:
   r2.1:
@@ -213,6 +234,24 @@ versions:
         - torchvision
       - tags: install,nccl,libs,_cuda
 
+  r4.0:
+    add_deps_recursive:
+      nvidia-inference-common-code:
+        version: r4.0
+      nvidia-scratch-space:
+        tags: _version.4_1
+    env:
+      BUILD_TRTLLM: 1
+    deps:
+      - tags: install,pytorch,from.src,_for-nvidia-mlperf-inference-v4.0
+        names:
+        - pytorch
+        - torch
+      - tags: install,torchvision,from.src,_for-nvidia-mlperf-inference-v4.0
+        names:
+        - pytorchvision
+        - torchvision
+
 docker:
   skip_run_cmd: 'no'
   all_gpus: 'yes'
diff --git a/script/get-git-repo/_cm.json b/script/get-git-repo/_cm.json
index 3d178f924..39fbcb1d2 100644
--- a/script/get-git-repo/_cm.json
+++ b/script/get-git-repo/_cm.json
@@ -17,6 +17,7 @@
     "folder": "CM_GIT_CHECKOUT_FOLDER",
     "patch": "CM_GIT_PATCH",
     "update": "CM_GIT_REPO_PULL",
+    "pull": "CM_GIT_REPO_PULL",
     "env_key": "CM_GIT_ENV_KEY",
     "submodules": "CM_GIT_RECURSE_SUBMODULES"
   },
diff --git a/script/get-ml-model-gptj/_cm.json b/script/get-ml-model-gptj/_cm.json
index f3a3f64b8..d7852c403 100644
--- a/script/get-ml-model-gptj/_cm.json
+++ b/script/get-ml-model-gptj/_cm.json
@@ -171,6 +171,13 @@
         }
       ]
     },
+    "fp8": {
+      "env": {
+        "CM_ML_MODEL_INPUT_DATA_TYPES": "fp8",
+        "CM_ML_MODEL_WEIGHT_DATA_TYPES": "fp8"
+      },
+      "group": "precision"
+    },
     "int4": {
       "env": {
         "CM_ML_MODEL_INPUT_DATA_TYPES": "int4",
@@ -193,6 +200,62 @@
       "group": "model-provider",
       "default": true
     },
+    "nvidia": {
+      "default_variations": {
+        "framework": "pytorch"
+      },
+      "group": "model-provider",
+      "env": {
+        "CM_TMP_ML_MODEL_PROVIDER": "nvidia"
+      }
+    },
+    "pytorch,nvidia": {
+      "default_variations": {
+        "precision": "fp8"
+      },
+      "deps": [
+        {
+          "tags": "get,git,repo,_repo.https://github.com/NVIDIA/TensorRT-LLM.git,_sha.0ab9d17a59c284d2de36889832fe9fc7c8697604",
+          "extra_cache_tags": "tensorrt-llm",
+          "env": {
+            "CM_GIT_CHECKOUT_PATH_ENV_NAME": "CM_TENSORRT_LLM_CHECKOUT_PATH"
+          }
+        },
+        {
+          "tags": "get,cuda",
+          "names": [
+            "cuda"
+          ]
+        },
+        {
+          "tags": "get,nvidia,scratch,space"
+        },
+        {
+          "tags": "get,cuda-devices"
+        },
+        {
+          "tags": "get,ml-model,gpt-j,_fp32,_pytorch",
+          "env": {
+          },
+          "force_new_env_keys": [
+            "GPTJ_CHECKPOINT_PATH"
+          ]
+        },
+        {
+          "tags": "get,nvidia,inference,common-code",
+          "names": [
+            "nvidia-inference-common-code"
+          ]
+        },
+        {
+          "tags": "get,python3",
+          "names": [
+            "python",
+            "python3"
+          ]
+        }
+      ]
+    },
     "intel": {
       "default_variations": {
         "framework": "pytorch"
diff --git a/script/get-ml-model-gptj/customize.py b/script/get-ml-model-gptj/customize.py
index d719f0095..4a22ff2ec 100644
--- a/script/get-ml-model-gptj/customize.py
+++ b/script/get-ml-model-gptj/customize.py
@@ -21,6 +21,12 @@ def preprocess(i):
             env['INT8_MODEL_DIR'] = os.getcwd()
         else:
             env['INT4_MODEL_DIR'] = os.getcwd()
+    elif env.get('CM_TMP_ML_MODEL_PROVIDER', '') == 'nvidia':
+        i['run_script_input']['script_name'] = 'run-nvidia'
+        gpu_arch = int(float(env['CM_CUDA_DEVICE_PROP_GPU_COMPUTE_CAPABILITY']) * 10)
+        env['CM_GPU_ARCH'] = gpu_arch
+        env['CM_TMP_REQUIRE_DOWNLOAD'] = 'no'
+
     else:
         is_saxml = env.get('CM_TMP_MODEL_SAXML','')
         if is_saxml == "fp32":
diff --git a/script/get-ml-model-gptj/run-nvidia.sh b/script/get-ml-model-gptj/run-nvidia.sh
new file mode 100644
index 000000000..b115f7838
--- /dev/null
+++ b/script/get-ml-model-gptj/run-nvidia.sh
@@ -0,0 +1,22 @@
+#!/bin/bash
+
+if [[ ! -e ${CM_NVIDIA_MLPERF_SCRATCH_PATH}/models/GPTJ-6B/checkpoint-final ]]; then
+  cp -r ${GPTJ_CHECKPOINT_PATH} ${CM_NVIDIA_MLPERF_SCRATCH_PATH}/models/GPTJ-6B/
+  test $? -eq 0 || exit $?
+fi
+
+echo "cd ${CM_TENSORRT_LLM_CHECKOUT_PATH}"
+cd ${CM_TENSORRT_LLM_CHECKOUT_PATH}
+
+make -C docker build
+test $? -eq 0 || exit $?
+
+RUN_CMD="bash -c '${CM_PYTHON_BIN_WITH_PATH} scripts/build_wheel.py -a=${CM_GPU_ARCH} --clean --install --trt_root /usr/local/tensorrt/ && python examples/quantization/quantize.py --dtype=float16  --output_dir=/mnt/models/GPTJ-6B/fp8-quantized-ammo/GPTJ-FP8-quantized --model_dir=/mnt/models/GPTJ-6B/checkpoint-final --qformat=fp8 --kv_cache_dtype=fp8 '"
+DOCKER_RUN_ARGS=" -v ${CM_NVIDIA_MLPERF_SCRATCH_PATH}:/mnt"
+export DOCKER_RUN_ARGS="$DOCKER_RUN_ARGS"
+export RUN_CMD="$RUN_CMD"
+make -C docker run LOCAL_USER=1
+test $? -eq 0 || exit $?
+
+${CM_PYTHON_BIN_WITH_PATH} ${CM_MLPERF_INFERENCE_NVIDIA_CODE_PATH}/code/gptj/tensorrt/onnx_tune.py --fp8-scalers-path=${CM_NVIDIA_MLPERF_SCRATCH_PATH}/models/GPTJ-6B/fp8-quantized-ammo/GPTJ-FP8-quantized/rank0.safetensors --scaler 1.005 --index 15
+test $? -eq 0 || exit $?
diff --git a/script/get-mlperf-inference-nvidia-common-code/_cm.json b/script/get-mlperf-inference-nvidia-common-code/_cm.json
index 46f12477e..7b47209ca 100644
--- a/script/get-mlperf-inference-nvidia-common-code/_cm.json
+++ b/script/get-mlperf-inference-nvidia-common-code/_cm.json
@@ -39,6 +39,9 @@
     },
     "nvidia-only": {
       "group": "repo-owner"
+    },
+    "go": {
+      "group": "repo-owner"
     }
   },
   "versions": {
@@ -62,6 +65,13 @@
           "version": "v3.1"
         }
       }
+    },
+    "r4.0": {
+      "add_deps_recursive": {
+        "mlperf-inference-results": {
+          "version": "v4.0"
+        }
+      }
     }
   }
 }
diff --git a/script/get-mlperf-inference-results/_cm.json b/script/get-mlperf-inference-results/_cm.json
index 46feecd04..f4870d57f 100644
--- a/script/get-mlperf-inference-results/_cm.json
+++ b/script/get-mlperf-inference-results/_cm.json
@@ -57,6 +57,12 @@
         "CM_MLPERF_INFERENCE_RESULTS_VERSION_NAME": "v3.1",
         "CM_GIT_URL": "https://github.com/<<<GITHUB_REPO_OWNER>>>/inference_results_v3.1.git"
       }
+    },
+    "v4.0": {
+      "env": {
+        "CM_MLPERF_INFERENCE_RESULTS_VERSION_NAME": "v4.0",
+        "CM_GIT_URL": "https://github.com/<<<GITHUB_REPO_OWNER>>>/inference_results_v4.0.git"
+      }
     }
   },
   "variations": {
@@ -85,6 +91,12 @@
         "GITHUB_REPO_OWNER": "GATEOverflow",
         "NVIDIA_ONLY": "yes"
       }
+    },
+    "go": {
+      "group": "source-repo",
+      "env": {
+        "GITHUB_REPO_OWNER": "GATEOverflow"
+      }
     }
   }
 }
diff --git a/script/install-ipex-from-src/_cm.json b/script/install-ipex-from-src/_cm.json
index f7bb3d675..630007d4f 100644
--- a/script/install-ipex-from-src/_cm.json
+++ b/script/install-ipex-from-src/_cm.json
@@ -124,6 +124,7 @@
             "conda-package",
             "setuptools"
           ],
+          "version": "69.5.1",
           "tags": "get,generic,conda-package,_package.setuptools,_source.conda-forge"
         },
         {
diff --git a/script/install-llvm-src/_cm.json b/script/install-llvm-src/_cm.json
index 42e7e20df..f15d4059c 100644
--- a/script/install-llvm-src/_cm.json
+++ b/script/install-llvm-src/_cm.json
@@ -299,7 +299,9 @@
            "tags": "get,generic-python-lib,_custom-python,_package.setuptools",
            "env": {
              "CM_PYTHON_BIN_WITH_PATH": "<<<CM_CONDA_BIN_PATH>>>/python3"
-           }
+           },
+           "version_max": "69.9.999",
+           "version_max_usable": "58.2.0"
         },
         {
            "tags": "get,generic-python-lib,_custom-python,_package.neural-compressor,_url.git+https://github.com/intel/neural-compressor.git@a2931eaa4052eec195be3c79a13f7bfa23e54473",
diff --git a/script/install-pytorch-from-src/_cm.json b/script/install-pytorch-from-src/_cm.json
index 057dbe681..67441d2bf 100644
--- a/script/install-pytorch-from-src/_cm.json
+++ b/script/install-pytorch-from-src/_cm.json
@@ -202,7 +202,8 @@
             "conda-package",
             "setuptools"
           ],
-          "tags": "get,generic,conda-package,_package.setuptools,_source.conda-forge"
+          "tags": "get,generic,conda-package,_package.setuptools,_source.conda-forge",
+          "version": "69.5.1"
         },
         {
           "names": [
@@ -220,6 +221,23 @@
         }
       ]
     },
+    "for-nvidia-mlperf-inference-v4.0": {
+      "base": [
+        "sha.32f93b1",
+        "cuda"
+      ],
+      "deps": [
+        {
+          "tags": "get,cmake",
+          "version_min": "3.25.0"
+	      }
+      ],
+      "ad": {
+        "pytorch-src-repo": {
+          "tags": "_no-recurse-submodules,_full-history"
+        }
+      }
+    },
     "for-nvidia-mlperf-inference-v3.1": {
       "base": [
         "sha.b5021ba9",
diff --git a/script/install-tensorflow-from-src/_cm.json b/script/install-tensorflow-from-src/_cm.json
index da28df774..059a62863 100644
--- a/script/install-tensorflow-from-src/_cm.json
+++ b/script/install-tensorflow-from-src/_cm.json
@@ -65,13 +65,15 @@
       "deps": [
         {
           "tags": "get,python",
-          "version_max": "3.10.999",
-          "version_min": "3.7.0",
+          "version_max": "3.12.999",
+          "version_min": "3.9.0",
           "names":["python","python3"]
         },
         {
-          "tags": "gcc,get",
-          "version_min": "9.3"
+          "tags": "get,llvm",
+          "version_min": "17.0.6",
+          "version_max": "17.9.99",
+          "version_max_usable": "17.0.6"
         },
         {
           "tags": "get,bazel",
@@ -333,6 +335,165 @@
       "env": {
         "CM_GIT_CHECKOUT": "v2.9.0"
       }
+    },
+    "v2.10.0": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.10.999",
+          "version_max_usable": "3.10.12",
+          "version_min": "3.7.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "gcc,get",
+          "version_max": "10",
+          "version_min": "9"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "5.1.1"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.10.0"
+      }
+    },
+    "v2.11.0": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.10.999",
+          "version_max_usable": "3.10.12",
+          "version_min": "3.7.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "gcc,get",
+          "version_max": "10",
+          "version_min": "9"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "5.3.0"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.11.0"
+      }
+    },
+    "v2.12.0": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.11.999",
+          "version_max_usable": "3.11.12",
+          "version_min": "3.7.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "gcc,get",
+          "version_max": "12",
+          "version_min": "9"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "5.3.0"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.12.0"
+      }
+    },
+    "v2.13.0": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.11.999",
+          "version_max_usable": "3.11.12",
+          "version_min": "3.8.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "get,llvm",
+          "version": "16.0.0"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "5.3.0"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.13.0"
+      }
+    },
+    "v2.14.0": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.11.999",
+          "version_max_usable": "3.11.12",
+          "version_min": "3.9.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "get,llvm",
+          "version": "16.0.0"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "6.1.0"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.14.0"
+      }
+    },
+    "v2.15.0": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.11.999",
+          "version_max_usable": "3.11.12",
+          "version_min": "3.9.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "get,llvm",
+          "version": "16.0.0"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "6.1.0"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.15.0"
+      }
+    },
+    "v2.16.1": {
+      "deps": [
+        {
+          "tags": "get,python",
+          "version_max": "3.12.999",
+          "version_max_usable": "3.11.12",
+          "version_min": "3.9.0",
+          "names":["python","python3"]
+        },
+        {
+          "tags": "get,llvm",
+          "version_min": "16.0.0",
+          "version_max_usable": "17.0.6",
+          "version_max": "17.9.999"
+        },
+        {
+          "tags": "get,bazel",
+          "version": "6.5.0"
+        }
+      ],
+      "env": {
+        "CM_GIT_CHECKOUT": "v2.16.1"
+      }
     }
   }
 }
diff --git a/script/install-tensorflow-from-src/customize.py b/script/install-tensorflow-from-src/customize.py
index 646c4f7de..3ddee9a90 100644
--- a/script/install-tensorflow-from-src/customize.py
+++ b/script/install-tensorflow-from-src/customize.py
@@ -6,6 +6,10 @@ def preprocess(i):
     if os_info['platform'] == 'windows':
         return {'return':1, 'error': 'Windows is not supported in this script yet'}
 
+    env = i['env']
+
+    env['CC'] = env['CM_C_COMPILER_WITH_PATH']
+
     return {'return':0}
 
 def postprocess(i):
diff --git a/script/install-torchvision-from-src/_cm.json b/script/install-torchvision-from-src/_cm.json
index a474bb484..f0e72134b 100644
--- a/script/install-torchvision-from-src/_cm.json
+++ b/script/install-torchvision-from-src/_cm.json
@@ -94,6 +94,19 @@
         "TORCH_CXX_FLAGS": "-D_GLIBCXX_USE_CXX11_ABI=1"
       }
     },
+    "for-nvidia-mlperf-inference-v4.0": {
+      "base": [
+        "sha.657027f3",
+        "cuda"
+      ],
+      "deps": [
+        {
+          "tags": "install,pytorch,from.src,_for-nvidia-mlperf-inference-v4.0"
+        }
+      ],
+      "env": {
+      }
+    },
     "for-nvidia-mlperf-inference-v3.1": {
       "base": [
         "sha.657027f3",
@@ -102,7 +115,7 @@
       "deps": [
         {
           "tags": "install,pytorch,from.src,_for-nvidia-mlperf-inference-v3.1"
-	}
+        }
       ],
       "env": {
       }
diff --git a/script/install-tpp-pytorch-extension/_cm.json b/script/install-tpp-pytorch-extension/_cm.json
index 4b3ce8659..83cf74ed3 100644
--- a/script/install-tpp-pytorch-extension/_cm.json
+++ b/script/install-tpp-pytorch-extension/_cm.json
@@ -120,6 +120,7 @@
             "conda-package",
             "setuptools"
           ],
+          "version": "69.5.1",
           "tags": "get,generic,conda-package,_package.setuptools,_source.conda-forge"
         },
         {
diff --git a/script/process-mlperf-accuracy/customize.py b/script/process-mlperf-accuracy/customize.py
index f7b13c16d..895227b44 100644
--- a/script/process-mlperf-accuracy/customize.py
+++ b/script/process-mlperf-accuracy/customize.py
@@ -25,7 +25,7 @@ def preprocess(i):
 
     results_dir_split = results_dir.split(xsep)
     dataset = env['CM_DATASET']
-    regenerate_accuracy_file = env.get('CM_MLPERF_REGENERATE_ACCURACY_FILE', False)
+    regenerate_accuracy_file = env.get('CM_MLPERF_REGENERATE_ACCURACY_FILE', env.get('CM_RERUN', False))
 
     for result_dir in results_dir_split:
 
@@ -105,9 +105,7 @@ def preprocess(i):
         else:
             return {'return': 1, 'error': 'Unsupported dataset'}
 
-        outfile = os.path.join(result_dir, "accuracy.txt")
-        if not os.path.exists(outfile) or (os.stat(outfile).st_size == 0) or env.get("CM_REGENERATE_MEASURE_FILES", False):
-            run_cmds.append(CMD)
+        run_cmds.append(CMD)
 
 
     if os_info['platform'] == 'windows':
diff --git a/script/run-mlperf-inference-app/_cm.yaml b/script/run-mlperf-inference-app/_cm.yaml
index b0a82a344..be61dae5c 100644
--- a/script/run-mlperf-inference-app/_cm.yaml
+++ b/script/run-mlperf-inference-app/_cm.yaml
@@ -37,6 +37,7 @@ default_env:
 
 input_mapping:
   backend: CM_MLPERF_BACKEND
+  beam_size: GPTJ_BEAM_SIZE
   category: CM_MLPERF_SUBMISSION_SYSTEM_TYPE
   clean: CM_MLPERF_CLEAN_ALL
   compliance: CM_MLPERF_LOADGEN_COMPLIANCE
@@ -253,6 +254,12 @@ variations:
       CM_RUN_MLPERF_INFERENCE_APP_DEFAULTS: r4.0_default
     group: benchmark-version
 
+  r4.1:
+    env:
+      CM_MLPERF_INFERENCE_VERSION: '4.1'
+      CM_RUN_MLPERF_INFERENCE_APP_DEFAULTS: r4.1_default
+    group: benchmark-version
+
   short:
     add_deps_recursive:
       submission-checker:
diff --git a/script/run-mlperf-inference-app/customize.py b/script/run-mlperf-inference-app/customize.py
index 4095eeb2c..861207c29 100644
--- a/script/run-mlperf-inference-app/customize.py
+++ b/script/run-mlperf-inference-app/customize.py
@@ -244,8 +244,7 @@ def preprocess(i):
     if state.get("cm-mlperf-inference-results"):
         #print(state["cm-mlperf-inference-results"])
         for sut in state["cm-mlperf-inference-results"]:#only one sut will be there
-            # Grigori: that may not work properly since customize may have another Python than MLPerf
-            # (for example, if we use virtual env)
+            # Better to do this in a stand alone CM script with proper deps but currently we manage this by modifying the sys path of the python executing CM
             import mlperf_utils
 
             print(sut)