| """ | |
| Unit tests for MultiModeBenchmarkRunner across diverse inference modes. | |
| """ | |
| import os | |
| import shutil | |
| import tempfile | |
| import unittest | |
| from framework.core.multi_mode_benchmark import MultiModeBenchmarkRunner | |
| class TestMultiModeBenchmark(unittest.TestCase): | |
| """Test suite verifying MultiModeBenchmarkRunner mode execution and metric reporting.""" | |
| def setUp(self): | |
| self.temp_dir = tempfile.mkdtemp() | |
| def tearDown(self): | |
| if os.path.exists(self.temp_dir): | |
| shutil.rmtree(self.temp_dir) | |
| def test_list_supported_modes(self): | |
| modes = MultiModeBenchmarkRunner.list_supported_modes() | |
| self.assertGreaterEqual(len(modes), 10) | |
| self.assertIn("dense_fp16", modes) | |
| self.assertIn("svd_linear_lowrank", modes) | |
| self.assertIn("rotational_slicegpt", modes) | |
| self.assertIn("mamba_ssm_scan", modes) | |
| def test_benchmark_attention_module(self): | |
| runner = MultiModeBenchmarkRunner( | |
| model_name_or_path="distilbert-base-uncased", | |
| device="cpu", | |
| batch_size=2, | |
| seq_len=16, | |
| num_samples=8, | |
| output_dir=self.temp_dir, | |
| ) | |
| res = runner.benchmark_attention_module( | |
| attention_type="svd_linear", | |
| hidden_size=32, | |
| num_heads=2, | |
| seq_len=16, | |
| batch_size=2, | |
| num_iters=3, | |
| ) | |
| self.assertIn("latency_ms", res) | |
| self.assertIn("peak_vram_mb", res) | |
| self.assertGreater(res["latency_ms"], 0) | |
| if __name__ == "__main__": | |
| unittest.main() | |
Xet Storage Details
- Size:
- 1.58 kB
- Xet hash:
- 7063789d67751076d195236ad630d10b3f99ba8c5f35b229b733140b31d6ddb7
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.