[Eagle] Remove the greedy branch and some redundant code (#4363)

Co-authored-by: Sehoon Kim <sehoon@x.ai>
2025-03-16 02:48:55 -07:00
parent 9971dc2283
commit 1b859295f4
14 changed files with 380 additions and 672 deletions
--- a/test/srt/test_eagle_infer.py
+++ b/test/srt/test_eagle_infer.py
@@ -122,8 +122,8 @@ class TestEAGLEEngine(unittest.TestCase):

    def _test_acc_length(self, engine):
        prompt = [
-            "Human: Give me a fully functional FastAPI server. Show the python code.\n\nAssistant:"
-        ] * 5
+            "Human: Give me a fully functional FastAPI server. Show the python code.\n\nAssistant:",
+        ] * 5  # test batched generation
        sampling_params = {"temperature": 0, "max_new_tokens": 512}
        output = engine.generate(prompt, sampling_params)
        output = output[0]