diff --git a/models/nlp/language_model/deberta/ixrt/README.md b/models/nlp/language_model/deberta/ixrt/README.md
index 56c3607e114fc5300eec0cd7e46c1a660ae7d430..1f9d1a28ac1938a170185fbe0d7d345b840d3952 100644
--- a/models/nlp/language_model/deberta/ixrt/README.md
+++ b/models/nlp/language_model/deberta/ixrt/README.md
@@ -38,8 +38,9 @@ bash scripts/prepare_model_and_dataset.sh
 ### Model Conversion
 
 ```bash
-wget https://raw.githubusercontent.com/bytedance/ByteMLPerf/main/byte_infer_perf/general_perf/model_zoo/deberta-torch-fp32.json
-python3 torch2onnx.py --model_path ./general_perf/model_zoo/popular/open_deberta/deberta-base-squad.pt --output_path deberta-torch-fp32.onnx
+tar -xvf open_deberta.tar
+wget <https://raw.githubusercontent.com/bytedance/ByteMLPerf/main/byte_infer_perf/general_perf/model_zoo/deberta-torch-fp32.json >
+python3 torch2onnx.py --model_path deberta-base-squad.pt --output_path deberta-torch-fp32.onnx
 onnxsim deberta-torch-fp32.onnx deberta-torch-fp32-sim.onnx
 python3 remove_clip_and_cast.py
 
@@ -62,36 +63,34 @@ bash scripts/infer_deberta_fp16_performance.sh
 
 ### Accuracy
 
-If you want to evaluate the accuracy of this model, please visit the website: < <https://github.com/yudefu/ByteMLPerf/tree/iluvatar_general_infer> >, which integrates inference and training of many models under this framework, supporting the ILUVATAR backend
+If you want to evaluate the accuracy of this model, please visit here: <toolbox/ByteMLPerf/tree/iluvatar_general_infer>, which integrates inference and training of many models under this framework, supporting the ILUVATAR backend
 
-For detailed steps regarding this model, please refer to this document: < <https://github.com/yudefu/ByteMLPerf/blob/iluvatar_general_infer/byte_infer_perf/general_perf/backends/ILUVATAR/README.zh_CN.md> > Note: You need to modify the relevant paths in the code to your own correct paths.
 
-```bash
-# clone and install requirements
-git clone https://github.com/yudefu/ByteMLPerf.git -b iluvatar_general_infer
-pip3 install -r ./ByteMLPerf/byte_infer_perf/general_perf/requirements.txt
-pip3 install -r ./ByteMLPerf/byte_infer_perf/general_perf/backends/ILUVATAR/requirements.txt
-
-# setup
-mv perf_engine.py ./ByteMLPerf/byte_infer_perf/general_perf/core/perf_engine.py
-cp ./datasets/open_squad/* ./ByteMLPerf/byte_infer_perf/general_perf/datasets/open_squad/
+For detailed steps regarding this model, please refer to this document: <toolbox/ByteMLPerf/blob/iluvatar_general_infer/byte_infer_perf/general_perf/backends/ILUVATAR/README.zh_CN.md> Note: You need to modify the relevant paths in the code to your own correct paths.
 
 mv ./deberta-sim-drop-clip-drop-invaild-cast.onnx general_perf/model_zoo/popular/open_deberta/
 mv ./general_perf/model_zoo/popular/ ./ByteMLPerf/byte_infer_perf/general_perf/model_zoo/
 
-cd /ByteMLPerf/byte_infer_perf/general_perf
-wget http://files.deepspark.org.cn:880/deepspark/Palak.tar
+pip3 install -r toolbox/ByteMLPerf/blob/iluvatar_general_infer/byte_infer_perf/general_perf/requirements.txt
+mv /ixrt/perf_engine.py toolbox/ByteMLPerf/byte_infer_perf/general_perf/core/perf_engine.py
+sftp -P 29880 vipzjtd@iftp.iluvatar.com.cn     密码：123..com
+get /upload/3-app/byteperf/Palak.tar
+exit
 tar -zxvf Palak.tar
 
-#接着修改代码：ByteMLPerf/byte_infer_perf/general_perf/datasets/open_squad/data_loader.py -AutoTokenizer.from_pretrained("Palak/microsoft_deberta-base_squad") => AutoTokenizer.from_pretrained("/Your/Path/Palak/microsoft_deberta-base_squad")
+接着修改代码：toolbox/ByteMLPerf/byte_infer_perf/general_perf/datasets/open_squad/data_loader.py
+AutoTokenizer.from_pretrained("Palak/microsoft_deberta-base_squad") => AutoTokenizer.from_pretrained("/Your/Path/Palak/microsoft_deberta-base_squad")
 
-# run acc perf
-sed -i 's/tensorrt_legacy/tensorrt/g' backends/ILUVATAR/common.py
+mv deberta-sim-drop-clip-drop-invaild-cast.onnx general_perf/model_zoo/popular/open_deberta/
+cd toolbox/ByteMLPerf/byte_infer_perf/
+mv /general_perf/general_perf/model_zoo/popular/open_deberta /general_perf/model_zoo/popular/open_deberta
+cd toolbox/ByteMLPerf/byte_infer_perf/general_perf
 python3 core/perf_engine.py --hardware_type ILUVATAR --task deberta-torch-fp32
 ```
 
+
 ## Results
 
-| Model   | BatchSize | Precision | QPS   | Exact Match | F1 Score |
-| ------- | --------- | --------- | ----- | ----------- | -------- |
-| DeBERTa | 1         | FP16      | 18.58 | 73.76       | 81.24    |
+Model   |BatchSize  |Precision |QPS       |Exact Match  |F1 Score
+--------|-----------|----------|----------|-------------|------------
+DeBerta |    16     |   FP16   | 18.58    | 73.76       | 81.24