Skip to content

GitLab

Switch branch/tag
Find file Normal viewHistoryPermalink
test_eagle_infer_beta.py 3.52 KB
Newer Older
Liangsheng Yin's avatar
Beta spec-overlap for EAGLE (#11398)  
Liangsheng Yin committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125 
import unittest
from types import SimpleNamespace

from sglang.srt.utils import kill_process_tree
from sglang.test.few_shot_gsm8k import run_eval
from sglang.test.test_utils import (
    DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
    DEFAULT_URL_FOR_TEST,
    CustomTestCase,
    popen_launch_server,
)


class TestEagleBS1(CustomTestCase):
    num_questions = 60

    @classmethod
    def setUpClass(cls):
        cls.model = "meta-llama/Llama-2-7b-chat-hf"
        cls.base_url = DEFAULT_URL_FOR_TEST
        cls.process = popen_launch_server(
            cls.model,
            cls.base_url,
            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
            other_args=[
                "--trust-remote-code",
                "--attention-backend",
                "triton",
                "--enable-beta-spec",
                "--speculative-algorithm",
                "EAGLE",
                "--speculative-draft-model",
                "lmzheng/sglang-EAGLE-llama2-chat-7B",
                "--speculative-num-steps",
                "5",
                "--speculative-eagle-topk",
                "1",
                "--speculative-num-draft-tokens",
                "6",
                "--max-running-requests",
                "1",
            ],
        )

    @classmethod
    def tearDownClass(cls):
        kill_process_tree(cls.process.pid)

    def test_gsm8k(self):
        args = SimpleNamespace(
            num_shots=5,
            data_path=None,
            num_questions=self.num_questions,
            max_new_tokens=512,
            parallel=128,
            host="http://127.0.0.1",
            port=int(self.base_url.split(":")[-1]),
        )
        metrics = run_eval(args)
        print(f"TestEagleBS1 -- {metrics=}")
        self.assertGreater(
            metrics["accuracy"], 0.33
        )  # 0.3333 for 60 questions; 0.234 for 1319 questions


class TestEagleLargeBS(CustomTestCase):
    num_questions = 10000
    max_running_requests = 64
    other_args = [
        "--trust-remote-code",
        "--attention-backend",
        "triton",
        "--enable-beta-spec",
        "--speculative-algorithm",
        "EAGLE",
        "--speculative-draft-model",
        "lmzheng/sglang-EAGLE-llama2-chat-7B",
        "--speculative-num-steps",
        "5",
        "--speculative-eagle-topk",
        "1",
        "--speculative-num-draft-tokens",
        "6",
        "--mem-fraction-static",
        "0.75",
        "--max-running-requests",
        str(max_running_requests),
        "--cuda-graph-bs",
        *[str(i) for i in range(1, max_running_requests + 1)],
    ]

    @classmethod
    def setUpClass(cls):
        cls.model = "meta-llama/Llama-2-7b-chat-hf"
        cls.base_url = DEFAULT_URL_FOR_TEST
        cls.process = popen_launch_server(
            cls.model,
            cls.base_url,
            timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
            other_args=cls.other_args,
        )

    @classmethod
    def tearDownClass(cls):
        kill_process_tree(cls.process.pid)

    def test_gsm8k(self):
        args = SimpleNamespace(
            num_shots=5,
            data_path=None,
            num_questions=self.num_questions,
            max_new_tokens=512,
            parallel=128,
            host="http://127.0.0.1",
            port=int(self.base_url.split(":")[-1]),
        )
        metrics = run_eval(args)
        print(f"TestEagleLargeBS -- {metrics=}")
        self.assertGreater(
            metrics["accuracy"], 0.23
        )  # 0.3333 for 60 questions; 0.234 for 1319 questions


if __name__ == "__main__":
    unittest.main()