Eagle speculative decoding part 4: Add EAGLE2 worker (#2150)
Co-authored-by: kavioyu <kavioyu@tencent.com> Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
This commit is contained in:
@@ -13,6 +13,7 @@ suites = {
|
||||
"test_abort.py",
|
||||
"test_chunked_prefill.py",
|
||||
"test_double_sparsity.py",
|
||||
"test_eagle_infer.py",
|
||||
"test_embedding_openai_server.py",
|
||||
"test_eval_accuracy_mini.py",
|
||||
"test_get_weights_by_name.py",
|
||||
|
||||
39
test/srt/test_eagle_infer.py
Normal file
39
test/srt/test_eagle_infer.py
Normal file
@@ -0,0 +1,39 @@
|
||||
import unittest
|
||||
|
||||
import sglang as sgl
|
||||
|
||||
|
||||
class TestEAGLEEngine(unittest.TestCase):
|
||||
|
||||
def test_eagle_accuracy(self):
|
||||
prompt = "Today is a sunny day and I like"
|
||||
target_model_path = "meta-llama/Llama-2-7b-chat-hf"
|
||||
speculative_draft_model_path = "lmzheng/sglang-EAGLE-llama2-chat-7B"
|
||||
|
||||
sampling_params = {"temperature": 0, "max_new_tokens": 8}
|
||||
|
||||
engine = sgl.Engine(
|
||||
model_path=target_model_path,
|
||||
speculative_draft_model_path=speculative_draft_model_path,
|
||||
speculative_algorithm="EAGLE",
|
||||
speculative_num_steps=3,
|
||||
speculative_eagle_topk=4,
|
||||
speculative_num_draft_tokens=16,
|
||||
)
|
||||
out1 = engine.generate(prompt, sampling_params)["text"]
|
||||
engine.shutdown()
|
||||
|
||||
engine = sgl.Engine(model_path=target_model_path)
|
||||
out2 = engine.generate(prompt, sampling_params)["text"]
|
||||
engine.shutdown()
|
||||
|
||||
print("==== Answer 1 ====")
|
||||
print(out1)
|
||||
|
||||
print("==== Answer 2 ====")
|
||||
print(out2)
|
||||
self.assertEqual(out1, out2)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
Reference in New Issue
Block a user