[CI] recover e2e test (#2688)

1. recover the skipped test.
2. remove pangu eager mode test, it's tested by torchair mode already.
3. skip pangu test util the bug is fixed.

- vLLM version: v0.10.1.1
- vLLM main:
56d04089ef

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-09-02 18:49:17 +08:00
committed by GitHub
parent f023bd52bf
commit 0829b4873f
6 changed files with 8 additions and 45 deletions

View File

@@ -6,7 +6,6 @@ import pytest
from tests.e2e.conftest import VllmRunner
from tests.e2e.model_utils import check_outputs_equal
from vllm_ascend.ascend_config import clear_ascend_config
MODELS = [
# for MHA
@@ -103,8 +102,6 @@ def test_prefix_cache_with_ascend_scheduler(model: str,
gpu_memory_utilization=0.7) as vllm_model:
vllm_output = vllm_model.generate_greedy(INPUT_PROMPTS, max_tokens)
clear_ascend_config()
with VllmRunner(model,
additional_config={
'ascend_scheduler_config': {
@@ -119,8 +116,6 @@ def test_prefix_cache_with_ascend_scheduler(model: str,
prefix_cache_output = vllm_model.generate_greedy(
INPUT_PROMPTS, max_tokens)
clear_ascend_config()
with VllmRunner(model,
additional_config={
'ascend_scheduler_config': {
@@ -136,8 +131,6 @@ def test_prefix_cache_with_ascend_scheduler(model: str,
chunk_prefill_prefix_cache_output = vllm_model.generate_greedy(
INPUT_PROMPTS, max_tokens)
clear_ascend_config()
check_outputs_equal(
outputs_0_lst=vllm_output,
outputs_1_lst=prefix_cache_output,