From cd8a89ef7a444e156b2f932202bc9265eee5b496 Mon Sep 17 00:00:00 2001
From: WG <39621324+wangguan1995@users.noreply.github.com>
Date: Fri, 16 Jun 2023 15:35:22 +0800
Subject: [PATCH] Tipc fix 0616 (#374)

* fix wrong reponame

* enlarge eularbeam epochs
---
 test_tipc/README.MD                                     | 2 +-
 test_tipc/benchmark_train.sh                            | 3 ++-
 test_tipc/configs/train_eular_beam/train_eular_beam.txt | 2 +-
 3 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/test_tipc/README.MD b/test_tipc/README.MD
index 2eaa2d1a1..52cb42c01 100644
--- a/test_tipc/README.MD
+++ b/test_tipc/README.MD
@@ -4,7 +4,7 @@ TIPC功能测试的主程序为benchmark_train.sh，可以测试基于Python的
 - docker image: registry.baidubce.com/paddlepaddle/paddle:latest-dev-cuda11.2-cudnn8-gcc82
 - paddle = 0.0.0
 - paddlescience = 0.0.0
-- python = 3.7
+- python = 3.10
 ## 2. 测试结论汇总
 | 算法名称 | 模型名称 | 单机单卡 | 单机多卡 | 多机多卡 | 模型压缩（单机多卡） |
 |  :----  |   :----  |    :----  |  :----   |  :----   |  :----   |
diff --git a/test_tipc/benchmark_train.sh b/test_tipc/benchmark_train.sh
index 15d3fda8b..90b6a6e5a 100644
--- a/test_tipc/benchmark_train.sh
+++ b/test_tipc/benchmark_train.sh
@@ -84,7 +84,6 @@ IFS=$'\n'
 lines=(${dataline})
 model_name=$(func_parser_value "${lines[1]}")
 workdir=$(func_parser_value "${lines[60]}")
-cd ${PDSC_DIR}/${workdir}
 python=$(func_parser_value "${lines[2]}")
 line_num=`grep -n "train_benchmark_params" $FILENAME  | cut -d ":" -f 1`
 batch_size=$(func_parser_value "${lines[line_num]}")
@@ -134,6 +133,8 @@ func_sed_params "$FILENAME" "${line_eval_py}" "null"
 func_sed_params "$FILENAME" "${line_export_py}" "null"
 func_sed_params "$FILENAME" "${line_python}"  "$python"
 
+cd ${PDSC_DIR}/${workdir}
+
 if  [ ! -n "$PARAMS" ] ;then
     # PARAMS input is not a word.
     IFS="|"
diff --git a/test_tipc/configs/train_eular_beam/train_eular_beam.txt b/test_tipc/configs/train_eular_beam/train_eular_beam.txt
index 6f51c10ec..c6d200563 100644
--- a/test_tipc/configs/train_eular_beam/train_eular_beam.txt
+++ b/test_tipc/configs/train_eular_beam/train_eular_beam.txt
@@ -52,7 +52,7 @@ null:null
 ===========================train_benchmark_params==========================
 batch_size:1
 fp_items:fp32
-epoch:20
+epoch:10000
 --profiler_options:batch_range=[1,1];state=GPU;tracer_option=Default;profile_path=model.profile
 flags:FLAGS_eager_delete_tensor_gb=0.0;FLAGS_fraction_of_gpu_memory_to_use=0.98;FLAGS_conv_workspace_size_limit=0
 ===========================prepare_params==========================