test_generation_models.py 5.98 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
# Copyright 2023-2024 SGLang Team
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
14
15
16
"""
Usage:

Kiv Chen's avatar
Kiv Chen committed
17
18
19
To test a specific model locally:
1. Add it to ALL_MODELS, for example, `ModelCase("Qwen/Qwen2-1.5B")`
2. Run `ONLY_RUN=Qwen/Qwen2-1.5B python3 -m unittest test_generation_models.TestGenerationModels`
20
21
22
"""

import dataclasses
23
import multiprocessing as mp
24
import os
Kiv Chen's avatar
Kiv Chen committed
25
import random
26
import unittest
27
from typing import List
28
29
30

import torch

31
32
33
34
35
36
from sglang.test.runners import (
    DEFAULT_PROMPTS,
    HFRunner,
    SRTRunner,
    check_close_model_outputs,
)
37
from sglang.test.test_utils import CustomTestCase, is_in_ci
38

39

40
41
42
43
44
@dataclasses.dataclass
class ModelCase:
    model_path: str
    tp_size: int = 1
    prefill_tolerance: float = 5e-2
45
    decode_tolerance: float = 6e-2  # Increased to fix numerical error in issue #8614.
46
    rouge_l_tolerance: float = 1
47
    skip_long_prompt: bool = False
48
    trust_remote_code: bool = False
49
50


51
# Popular models that run on the CI
52
CI_MODELS = [
53
    ModelCase("meta-llama/Llama-3.1-8B-Instruct"),
54
    ModelCase("google/gemma-2-2b"),
55
]
56

Kiv Chen's avatar
Kiv Chen committed
57
58
59
# the complete set of models to test sglang's generation model
ALL_MODELS = [
    *CI_MODELS,
60
    ModelCase("Qwen/Qwen2-1.5B"),
61
    ModelCase("Qwen/Qwen2.5-14B-Instruct"),
62
63
    ModelCase("HuggingFaceTB/SmolLM-135M-Instruct", skip_long_prompt=True),
    ModelCase("allenai/OLMo-1B-0724-hf", decode_tolerance=8e-2, skip_long_prompt=True),
64
65
66
    ModelCase(
        "THUDM/glm-4-9b-chat", tp_size=2, trust_remote_code=True, skip_long_prompt=True
    ),
Chayenne's avatar
Chayenne committed
67
    ModelCase("openai-community/gpt2"),
68
    ModelCase("microsoft/phi-1_5", trust_remote_code=True),
69
    ModelCase("adept/persimmon-8b-chat"),
70
    ModelCase("inclusionAI/Ling-lite", trust_remote_code=True),
Kiv Chen's avatar
Kiv Chen committed
71
    ModelCase("microsoft/Phi-3-small-8k-instruct", trust_remote_code=True),
Jani Monoses's avatar
Jani Monoses committed
72
    ModelCase("allenai/OLMo-2-1124-7B-Instruct", skip_long_prompt=True),
73
    ModelCase("ibm-granite/granite-3.0-2b-instruct", skip_long_prompt=True),
74
75
76
77
78
79
    ModelCase(
        "microsoft/Phi-3.5-MoE-instruct",
        tp_size=2,
        trust_remote_code=True,
        skip_long_prompt=True,
    ),
wenhuipeng's avatar
wenhuipeng committed
80
    ModelCase("facebook/opt-125m", skip_long_prompt=True),
81
82
83
84
85
86
    ModelCase(
        "nvidia/Llama-3_3-Nemotron-Super-49B-v1_5",
        tp_size=2,
        trust_remote_code=True,
        skip_long_prompt=True,
    ),
87
88
89
90
91
92
    ModelCase(
        "nvidia/Llama-3_1-Nemotron-Ultra-253B-v1",
        tp_size=8,
        trust_remote_code=True,
        skip_long_prompt=True,
    ),
93
]
94

95
TORCH_DTYPES = [torch.float16]
96
97


98
class TestGenerationModels(CustomTestCase):
99

100
101
    @classmethod
    def setUpClass(cls):
102
        mp.set_start_method("spawn", force=True)
103

104
    def assert_close_logits_and_output_strs(
105
        self,
106
107
108
        prompts: List[str],
        model_case: ModelCase,
        torch_dtype: torch.dtype,
109
    ) -> None:
110
111
112
113
114
115
116
        model_path = model_case.model_path
        prefill_tolerance, decode_tolerance, rouge_l_tolerance = (
            model_case.prefill_tolerance,
            model_case.decode_tolerance,
            model_case.rouge_l_tolerance,
        )
        max_new_tokens = 32
117

118
        with HFRunner(
119
120
121
            model_path,
            torch_dtype=torch_dtype,
            model_type="generation",
122
            trust_remote_code=model_case.trust_remote_code,
123
        ) as hf_runner:
124
            hf_outputs = hf_runner.forward(prompts, max_new_tokens=max_new_tokens)
125
126
127

        with SRTRunner(
            model_path,
128
            tp_size=model_case.tp_size,
129
            torch_dtype=torch_dtype,
130
            model_type="generation",
131
            trust_remote_code=model_case.trust_remote_code,
132
        ) as srt_runner:
133
            srt_outputs = srt_runner.forward(prompts, max_new_tokens=max_new_tokens)
134

135
136
137
138
139
140
141
        check_close_model_outputs(
            hf_outputs=hf_outputs,
            srt_outputs=srt_outputs,
            prefill_tolerance=model_case.prefill_tolerance,
            decode_tolerance=model_case.decode_tolerance,
            rouge_l_tolerance=model_case.rouge_l_tolerance,
            debug_text=f"model_path={model_path} prompts={prompts}",
142
        )
143

Kiv Chen's avatar
Kiv Chen committed
144
    @unittest.skipIf(not is_in_ci(), "Local test should run all models")
145
146
    def test_ci_models(self):
        for model_case in CI_MODELS:
147
            for torch_dtype in TORCH_DTYPES:
Kiv Chen's avatar
Kiv Chen committed
148
                prompts = DEFAULT_PROMPTS
149
150
151
152
153
154

                # Skip long prompts for models that do not have a long context
                if model_case.skip_long_prompt:
                    prompts = [p for p in DEFAULT_PROMPTS if len(p) < 1000]

                # Assert the logits and output strs are close
155
                self.assert_close_logits_and_output_strs(
156
                    prompts, model_case, torch_dtype
157
158
                )

Kiv Chen's avatar
Kiv Chen committed
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
    @unittest.skipIf(is_in_ci(), "CI only runs selected models for simplicity")
    def test_all_models(self):
        for model_case in ALL_MODELS:
            for torch_dtype in TORCH_DTYPES:
                if (
                    "ONLY_RUN" in os.environ
                    and os.environ["ONLY_RUN"] != model_case.model_path
                ):
                    continue

                # Skip long prompts for models that do not have a long context
                prompts = DEFAULT_PROMPTS
                if model_case.skip_long_prompt:
                    prompts = [p for p in DEFAULT_PROMPTS if len(p) < 1000]

                # Assert the logits and output strs are close
                self.assert_close_logits_and_output_strs(
                    prompts, model_case, torch_dtype
                )
178

179

180
if __name__ == "__main__":
Mingyi's avatar
Mingyi committed
181
    unittest.main()