[VLM] Add PP support and fix GPTQ inference for Ovis models (#18958)

Signed-off-by: isotr0py <2037008807@qq.com> Signed-off-by: Isotr0py <2037008807@qq.com>
2025-05-31 01:11:44 +08:00
parent f49239cb45
commit 5a8641638a
5 changed files with 145 additions and 91 deletions
--- a/tests/distributed/test_pipeline_parallel.py
+++ b/tests/distributed/test_pipeline_parallel.py
@@ -227,6 +227,7 @@ MULTIMODAL_MODELS = {
    "llava-hf/llava-onevision-qwen2-0.5b-ov-hf": PPTestSettings.fast(),
    "openbmb/MiniCPM-Llama3-V-2_5": PPTestSettings.fast(),
    "allenai/Molmo-7B-D-0924": PPTestSettings.fast(),
+    "AIDC-AI/Ovis2-1B": PPTestSettings.fast(),
    "microsoft/Phi-3.5-vision-instruct": PPTestSettings.fast(),
    "mistralai/Pixtral-12B-2409": PPTestSettings.fast(load_format="dummy"),
    "Qwen/Qwen-VL-Chat": PPTestSettings.fast(),