jiangzhs/OpenCompass
1
0
Fork 0
mirror of https://github.com/open-compass/opencompass.git synced 2025-05-30 16:03:24 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[CI] Fix testcase failure (#1582)

Browse Source 
* update

* Update oc_score_baseline.yaml

* Update daily-run-test.yml

* Update daily-run-test.yml

* Update daily-run-test.yml

* Update daily-run-test.yml

---------

Co-authored-by: zhulin1 <zhulin1@pjlab.org.cn>
This commit is contained in:
zhulinJulia24 2024-10-02 12:30:38 +08:00 committed by GitHub
parent 22a4e76511
commit 89abcba486
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 11 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
.github/scripts/oc_score_assert.py vendored
View File

@ -7,8 +7,7 @@ import yaml
output_path = 'regression_result_daily'
chat_model_list = [
'baichuan2-7b-chat-hf', 'glm-4-9b-chat-turbomind', 'glm-4-9b-chat-vllm',
'deepseek-7b-chat-hf', 'deepseek-moe-16b-chat-hf',
'baichuan2-7b-chat-hf', 'deepseek-7b-chat-hf', 'deepseek-moe-16b-chat-hf',
'deepseek-v2-lite-chat-hf', 'deepseek-7b-chat-vllm', 'gemma2-2b-it-hf',
'gemma2-9b-it-hf', 'gemma-7b-it-vllm', 'internlm2_5-7b-chat-hf',
'internlm2_5-20b-chat-hf', 'internlm2_5-7b-chat-turbomind',

12
.github/scripts/oc_score_baseline.yaml vendored
View File

@ -244,14 +244,14 @@ gemma-7b-hf:
race-high: 66
gemma2-2b-hf:
gsm8k: 8
race-middle: 31
race-high: 30
gsm8k: 33
race-middle: 56
race-high: 58
gemma2-9b-hf:
gsm8k: 20
race-middle: 42
race-high: 35
gsm8k: 70
race-middle: 82
race-high: 84
internlm2_5-7b-hf:
gsm8k: 47

8
.github/workflows/daily-run-test.yml vendored
View File

@ -123,16 +123,16 @@ jobs:
conda info --envs
export from_tf=TRUE
python tools/list_configs.py internlm2_5 mmlu
opencompass --models hf_internlm2_5_7b --datasets race_ppl --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd1_${{ matrix.cuda_env }} --reuse --max-num-workers 2
opencompass --models hf_internlm2_5_7b --datasets race_ppl --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd1_${{ matrix.cuda_env }} --reuse
rm regression_result_daily -f && ln -s /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd1_${{ matrix.cuda_env }}/*/summary regression_result_daily
python -m pytest -m case1 -s -v --color=yes .github/scripts/oc_score_assert.py
opencompass --models hf_internlm2_5_7b_chat hf_internlm2_5_1_8b_chat --datasets race_gen -a lmdeploy --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd2 --reuse --max-num-workers 2
opencompass --models hf_internlm2_5_7b_chat --datasets race_gen -a lmdeploy --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd2_${{ matrix.cuda_env }} --reuse
rm regression_result_daily -f && ln -s /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd2_${{ matrix.cuda_env }}/*/summary regression_result_daily
python -m pytest -m case2 -s -v --color=yes .github/scripts/oc_score_assert.py
opencompass --datasets race_ppl --hf-type base --hf-path internlm/internlm2_5-7b --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd3 --reuse --max-num-workers 2
opencompass --datasets race_ppl --hf-type base --hf-path internlm/internlm2_5-7b --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd3_${{ matrix.cuda_env }} --reuse
rm regression_result_daily -f && ln -s /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd3_${{ matrix.cuda_env }}/*/summary regression_result_daily
python -m pytest -m case3 -s -v --color=yes .github/scripts/oc_score_assert.py
opencompass --datasets race_gen --hf-type chat --hf-path internlm/internlm2_5-7b-chat --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd4 --reuse --max-num-workers 2
opencompass --datasets race_gen --hf-type chat --hf-path internlm/internlm2_5-7b-chat --work-dir /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd4_${{ matrix.cuda_env }} --reuse
rm regression_result_daily -f && ln -s /cpfs01/user/qa-llm-cicd/report/${{ github.run_id }}/cmd4_${{ matrix.cuda_env }}/*/summary regression_result_daily
python -m pytest -m case4 -s -v --color=yes .github/scripts/oc_score_assert.py
- name: Remove Conda Env