sglangv0.5.2 & support Qwen3-Next-80B-A3B-Instruct

This commit is contained in:
maxiao1
2025-09-13 17:00:20 +08:00
commit 118f1fc726
2037 changed files with 515371 additions and 0 deletions

27
test/srt/test_gguf.py Normal file
View File

@@ -0,0 +1,27 @@
import unittest
from huggingface_hub import hf_hub_download
import sglang as sgl
from sglang.test.test_utils import CustomTestCase
class TestGGUF(CustomTestCase):
def test_models(self):
prompt = "Today is a sunny day and I like"
sampling_params = {"temperature": 0, "max_new_tokens": 8}
model_path = hf_hub_download(
"Qwen/Qwen2-1.5B-Instruct-GGUF",
filename="qwen2-1_5b-instruct-q4_k_m.gguf",
)
engine = sgl.Engine(model_path=model_path, random_seed=42, cuda_graph_max_bs=2)
outputs = engine.generate(prompt, sampling_params)["text"]
engine.shutdown()
self.assertEqual(outputs, " it. I have a lot of work")
if __name__ == "__main__":
unittest.main()