Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

fix vlm gguf ut (#895)

Heng Guo65b7ec2f6d48
diff --git a/test/test_cpu/test_gguf_format.py b/test/test_cpu/test_gguf_format.py
index 96b489ffe..d71920b39 100644
--- a/test/test_cpu/test_gguf_format.py
+++ b/test/test_cpu/test_gguf_format.py
@@ -329,7 +329,6 @@ class TestGGUF(unittest.TestCase):
autoround.quantize_and_save(output_dir=quantized_model_path, format="gguf:q4_0")
self.assertTrue("mmproj-model.gguf" in os.listdir("./saved"))
for file_name in os.listdir(quantized_model_path):
- file_name = os.listdir(quantized_model_path)[0]
file_size = os.path.getsize(os.path.join(quantized_model_path, file_name)) / 1024**2
if file_name == "mmproj-model.gguf":
self.assertAlmostEqual(file_size, 2535, delta=1.0)