Quality Assurance for v0.12.0

[XuehaoSun]

Problem Description

Quality Assurance for v0.12.0

[XuehaoSun]

cuda gpu test report: cuda unit test

• test_moe_model @lvliang-intel

[chensuyue]

intel xpu CI issue:

• test_xpu/test_autoround.py core dump, log1 , log2 , fixed by Grep core dumped issue in UT test #1484

[XuehaoSun]

test report: cuda unit test

• sglang_test_sglang @mengniwang95
• test_export @xin3he
• test_fp8_input @xin3he
• test_gguf @n1ck-guo
• test_mllm @xin3he
• test_transformers @lvliang-intel

[mengniwang95]

sglang fix: #1541

[XuehaoSun]

test report: cuda unit test

• test_alg_ext @xin3he
• test_export @xin3he
• test_fp8_input @xin3he
• test_gguf @n1ck-guo
• test_main_func @xin3he
• test_mllm @xin3he
• test_multiple_card @xin3he

[xin3he]

CUDA UT will use this change.
#1473

[XuehaoSun]

test report: cuda unit test

• test_llmc_integration @xin3he
• test_auto_gptq_format @xin3he
• test_auto_scheme @xin3he
• test_gguf_format @xin3he
• test_omni_model @xin3he fix: refactor and fix Omni model tests to improve CI stability #1598
• vllm_test_vllm @xin3he
• @WeiweiZhang1 test_cuda/integrations/test_vllm.py::test_auto_round_awq_format_vllm RuntimeError: expected scalar type Half but found BFloat16
• test_support_vlms @xin3he