feat: frequency, min_new_tokens, presence, and repetition penalties (#973)

test/srt/run_suite.py

@@ -11,9 +11,20 @@ suites = {
         "test_chunked_prefill.py",
         "test_torch_compile.py",
         "models/test_causal_models.py",
+        "sampling/penaltylib",
     ],
+    "sampling/penaltylib": glob.glob(
+        "sampling/penaltylib/**/test_*.py", recursive=True
+    ),
 }
 
+for target_suite_name, target_tests in suites.items():
+    for suite_name, tests in suites.items():
+        if suite_name == target_suite_name:
+            continue
+        if target_suite_name in tests:
+            tests.remove(target_suite_name)
+            tests.extend(target_tests)
 
 if __name__ == "__main__":
     arg_parser = argparse.ArgumentParser()