diff --git a/.github/workflows/llm_unit_tests.yml b/.github/workflows/llm_unit_tests.yml
index 674268c5..01155781 100644
--- a/.github/workflows/llm_unit_tests.yml
+++ b/.github/workflows/llm_unit_tests.yml
@@ -235,6 +235,9 @@ jobs:
           echo "MPT_7B_ORIGIN_PATH=${ORIGIN_DIR}/mpt-7b-chat" >> "$GITHUB_ENV"
           echo "WHISPER_TINY_ORIGIN_PATH=${ORIGIN_DIR}/whisper-tiny" >> "$GITHUB_ENV"
 
+          echo "MISTRAL_7B_INSTRUCT_V0_1_ORIGIN_PATH=${ORIGIN_DIR}/Mistral-7B-Instruct-v0.1" >> "$GITHUB_ENV"
+          echo "BAICHUAN2_7B_ORIGIN_PATH=${ORIGIN_DIR}/Baichuan2-7B-Chat" >> "$GITHUB_ENV"
+          echo "QWEN_7B_ORIGIN_PATH=${ORIGIN_DIR}/Qwen-7B-Chat" >> "$GITHUB_ENV"
       - name: Checkout repo
         uses: actions/checkout@f43a0e5ff2bd294095638e18286ca9a3d1956744 # actions/checkout@v3
 
@@ -303,6 +306,18 @@ jobs:
             echo "Directory $SPEECH_DATASET_PATH not found. Downloading from FTP server..."
             wget -r -nH --no-verbose --cut-dirs=2 $LLM_FTP_URL/llm/datasets/librispeech_asr_dummy -P $DATASET_DIR
           fi
+          if [ ! -d $MISTRAL_7B_INSTRUCT_V0_1_ORIGIN_PATH ]; then
+            echo "Directory $MISTRAL_7B_INSTRUCT_V0_1_ORIGIN_PATH not found. Downloading from FTP server..."
+            wget -r -nH --no-verbose --cut-dirs=1 $LLM_FTP_URL/llm/Mistral-7B-Instruct-v0.1 -P $ORIGIN_DIR
+          fi
+          if [ ! -d $QWEN_7B_ORIGIN_PATH ]; then
+            echo "Directory $QWEN_7B_ORIGIN_PATH not found. Downloading from FTP server..."
+            wget -r -nH --no-verbose --cut-dirs=1 $LLM_FTP_URL/llm/Qwen-7B-Chat -P $ORIGIN_DIR
+          fi
+          if [ ! -d $BAICHUAN2_7B_ORIGIN_PATH ]; then
+            echo "Directory $BAICHUAN2_7B_ORIGIN_PATH not found. Downloading from FTP server..."
+            wget -r -nH --no-verbose --cut-dirs=1 $LLM_FTP_URL/llm/Baichuan2-7B-Chat -P $ORIGIN_DIR
+          fi
           
       - name: Run LLM inference test
         shell: bash
@@ -313,8 +328,10 @@ jobs:
           elif [[ '${{ matrix.pytorch-version }}' == '2.0' ]]; then
             source /opt/intel/oneapi/setvars.sh
           fi
-          python -m pip install datasets librosa soundfile einops
+          python -m pip install datasets librosa soundfile einops tiktoken transformers_stream_generator
           bash python/llm/test/run-llm-inference-tests-gpu.sh
+          python -m pip install transformers==4.34.0 
+          bash python/llm/test/run-llm-inference-tests-gpu-434.sh
 
       - name: Run LLM example tests
         shell: bash
diff --git a/python/llm/test/inference_gpu/test_transformers_api_mlp.py b/python/llm/test/inference_gpu/test_transformers_api_mlp.py
new file mode 100644
index 00000000..4bba9c4c
--- /dev/null
+++ b/python/llm/test/inference_gpu/test_transformers_api_mlp.py
@@ -0,0 +1,129 @@
+#
+# Copyright 2016 The BigDL Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+ 
+import os
+import pytest
+ 
+import torch
+from bigdl.llm.transformers import AutoModelForCausalLM, AutoModel
+from transformers import LlamaTokenizer, AutoTokenizer
+ 
+device = os.environ['DEVICE']
+print(f'Running on {device}')
+ 
+PROMPT = "Once upon a time, there existed a little girl who liked to have adventures. She wanted to go to places and meet new people, and have fun"
+TEST_MODEL_LIST = [
+    ("Qwen-7B-Chat", AutoModelForCausalLM, AutoTokenizer, os.environ.get('QWEN_7B_ORIGIN_PATH')),
+    ("Mistral-7B-Instruct-v0.1", AutoModelForCausalLM, AutoTokenizer, os.environ.get('MISTRAL_7B_INSTRUCT_V0_1_ORIGIN_PATH'))
+]
+ 
+class Test_Optimize_Gpu_Model:
+    def setup_method(self):
+        self.layer_outputs = []
+        self.pre_layer_outputs = []
+ 
+    def run_optimize_gpu_model(self, Name, Model, Tokenizer, model_path, MLP_layer, layer_before_MLP, lower_bound):
+        with torch.inference_mode():
+            def pre_forward_hook(module, input, output, layer_name):
+                self.pre_layer_outputs.append(output)
+                
+            def forward_hook(module, input, output, layer_name):
+                self.layer_outputs.append(output)
+ 
+            tokenizer = Tokenizer.from_pretrained(model_path, trust_remote_code=True)
+            input_ids = tokenizer.encode(PROMPT, return_tensors="pt").to(device)
+ 
+            model = Model.from_pretrained(model_path,
+                                        load_in_4bit=True,
+                                        optimize_model=False,
+                                        trust_remote_code=True)
+            model = model.to(device)
+            for layer_name, layer_module in model.named_modules():
+                if layer_name == layer_before_MLP:
+                    layer_module.register_forward_hook(
+                        lambda module, input, output, layer_name=layer_name: pre_forward_hook(module, input,
+                                                                                            output, layer_name))
+                if layer_name == MLP_layer:
+                    layer_module.register_forward_hook(
+                        lambda module, input, output, layer_name=layer_name: forward_hook(module, input,
+                                                                                        output, layer_name))
+            logits_base_model = (model(input_ids)).logits
+            # the list `layer_output` has only one element.
+            layer_tensor = self.layer_outputs.pop()
+            model.to('cpu')
+ 
+            opt_model = Model.from_pretrained(model_path,
+                                            load_in_4bit=True,
+                                            optimize_model=True,
+                                            trust_remote_code=True)
+            opt_model = opt_model.to(device)
+ 
+ 
+            def replace_forward_hook(module, input, output, layer_name):
+                output = self.pre_layer_outputs[0]
+                return output
+ 
+            for layer_name, layer_module in opt_model.named_modules():
+                if layer_name == layer_before_MLP:
+                    layer_module.register_forward_hook(
+                        lambda module, input, output, layer_name=layer_name: replace_forward_hook(module, input,
+                                                                                                output, layer_name))
+                if layer_name == MLP_layer:
+                    layer_module.register_forward_hook(
+                        lambda module, input, output, layer_name=layer_name: forward_hook(module, input,
+                                                                                        output, layer_name))
+            logits_optimized_model = (opt_model(input_ids)).logits
+            # the list `layer_output` has only one element.
+            opt_layer_tensor = self.layer_outputs[0]
+            opt_model.to('cpu')
+ 
+ 
+            MLP_output_diff = []
+            for i, (t1, t2) in enumerate(zip(layer_tensor, opt_layer_tensor)):
+                if t1 is not None and t2 is not None:
+                    if isinstance(t1, torch.Tensor) and isinstance(t2, torch.Tensor):
+                        MLP_output_diff.append(t1 - t2)
+                    else:
+                        # 'past_key_value'is of type tuple as default.
+                        for i, (t3, t4) in enumerate(zip(t1, t2)):
+                            MLP_output_diff.append(t3 - t4)
+ 
+            max_diff_tensor = [torch.max(item).item() for item in MLP_output_diff]
+            print(max_diff_tensor)
+           
+            assert all(max_diff <= lower_bound for max_diff in max_diff_tensor)
+   
+    @pytest.mark.parametrize('Name, Model, Tokenizer, model_path',TEST_MODEL_LIST)
+    def test_dynamic_functions(self, Name, Model, Tokenizer, model_path):
+        if Name == "Qwen-7B-Chat":
+            self.Qwen_7B_gpu_model(Name, Model, Tokenizer, model_path)
+        elif Name == "Mistral-7B-Instruct-v0.1":
+            self.Mistral_7B_Instruct_gpu_model(Name, Model, Tokenizer, model_path)
+ 
+   
+    def Qwen_7B_gpu_model(self, Name, Model, Tokenizer, model_path):
+        # currently only compare the output of the last mlp layer.
+        layer_before_MLP = "transformer.h.31.ln_2"
+        MLP_layer = "transformer.h.31.mlp"
+        lower_bound = 0
+        self.run_optimize_gpu_model(Name, Model, Tokenizer, model_path, MLP_layer, layer_before_MLP, lower_bound)
+   
+    def Mistral_7B_Instruct_gpu_model(self, Name, Model, Tokenizer, model_path):
+        # currently only compare the output of the last mlp layer.
+        layer_before_MLP = "model.layers.31.post_attention_layernorm"
+        MLP_layer = "model.layers.31.mlp"
+        lower_bound = 0
+        self.run_optimize_gpu_model(Name, Model, Tokenizer, model_path, MLP_layer, layer_before_MLP, lower_bound)
\ No newline at end of file
diff --git a/python/llm/test/run-llm-inference-tests-gpu-434.sh b/python/llm/test/run-llm-inference-tests-gpu-434.sh
new file mode 100644
index 00000000..ea5fe8b0
--- /dev/null
+++ b/python/llm/test/run-llm-inference-tests-gpu-434.sh
@@ -0,0 +1,28 @@
+#!/bin/bash
+
+export ANALYTICS_ZOO_ROOT=${ANALYTICS_ZOO_ROOT}
+export LLM_HOME=${ANALYTICS_ZOO_ROOT}/python/llm/src
+export LLM_INFERENCE_TEST_DIR=${ANALYTICS_ZOO_ROOT}/python/llm/test/inference_gpu
+
+export USE_XETLA=OFF
+export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
+export DEVICE='xpu'
+
+set -e
+
+echo "# Start testing inference"
+start=$(date "+%s")
+
+if [ -z "$THREAD_NUM" ]; then
+  THREAD_NUM=2
+fi
+export OMP_NUM_THREADS=$THREAD_NUM
+export BIGDL_LLM_XMX_DISABLED=1
+pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_mlp.py -v -s -k "Mistral"
+unset BIGDL_LLM_XMX_DISABLED
+
+now=$(date "+%s")
+time=$((now-start))
+
+echo "Bigdl-llm gpu inference tests for transformers 4.34.0 finished"
+echo "Time used:$time seconds"
diff --git a/python/llm/test/run-llm-inference-tests-gpu.sh b/python/llm/test/run-llm-inference-tests-gpu.sh
index 6951ddc3..fc4b6f90 100644
--- a/python/llm/test/run-llm-inference-tests-gpu.sh
+++ b/python/llm/test/run-llm-inference-tests-gpu.sh
@@ -21,6 +21,7 @@ pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api.py -v -s
 export BIGDL_LLM_XMX_DISABLED=1
 pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_final_logits.py -v -s
 pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_attention.py -v -s
+pytest ${LLM_INFERENCE_TEST_DIR}/test_transformers_api_mlp.py -v -s -k "not Mistral"
 unset BIGDL_LLM_XMX_DISABLED
 
 now=$(date "+%s")