Skip to content

vllm.entrypoints.cli.openai

ChatCommand

Bases: CLISubcommand

The chat subcommand for the vLLM CLI.

Source code in vllm/entrypoints/cli/openai.py 
 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
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
class ChatCommand(CLISubcommand):
    """The `chat` subcommand for the vLLM CLI. """
    name = "chat"

    @staticmethod
    def cmd(args: argparse.Namespace) -> None:
        model_name, client = _interactive_cli(args)
        system_prompt = args.system_prompt
        conversation: list[ChatCompletionMessageParam] = []

        if system_prompt is not None:
            conversation.append({"role": "system", "content": system_prompt})

        if args.quick:
            conversation.append({"role": "user", "content": args.quick})

            chat_completion = client.chat.completions.create(
                model=model_name, messages=conversation)
            print(chat_completion.choices[0].message.content)
            return

        print("Please enter a message for the chat model:")
        while True:
            try:
                input_message = input("> ")
            except EOFError:
                return
            conversation.append({"role": "user", "content": input_message})

            chat_completion = client.chat.completions.create(
                model=model_name, messages=conversation)

            response_message = chat_completion.choices[0].message
            output = response_message.content

            conversation.append(response_message)  # type: ignore
            print(output)

    def subparser_init(
            self,
            subparsers: argparse._SubParsersAction) -> FlexibleArgumentParser:
        chat_parser = subparsers.add_parser(
            "chat",
            help="Generate chat completions via the running API server.",
            description="Generate chat completions via the running API server.",
            usage="vllm chat [options]")
        _add_query_options(chat_parser)
        chat_parser.add_argument(
            "--system-prompt",
            type=str,
            default=None,
            help=("The system prompt to be added to the chat template, "
                  "used for models that support system prompts."))
        chat_parser.add_argument("-q",
                                 "--quick",
                                 type=str,
                                 metavar="MESSAGE",
                                 help=("Send a single prompt as MESSAGE "
                                       "and print the response, then exit."))
        return chat_parser

name class-attribute instance-attribute 

name = 'chat'

cmd staticmethod 

cmd(args: Namespace) -> None
Source code in vllm/entrypoints/cli/openai.py 
 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
126
127
128
129
130
131
@staticmethod
def cmd(args: argparse.Namespace) -> None:
    model_name, client = _interactive_cli(args)
    system_prompt = args.system_prompt
    conversation: list[ChatCompletionMessageParam] = []

    if system_prompt is not None:
        conversation.append({"role": "system", "content": system_prompt})

    if args.quick:
        conversation.append({"role": "user", "content": args.quick})

        chat_completion = client.chat.completions.create(
            model=model_name, messages=conversation)
        print(chat_completion.choices[0].message.content)
        return

    print("Please enter a message for the chat model:")
    while True:
        try:
            input_message = input("> ")
        except EOFError:
            return
        conversation.append({"role": "user", "content": input_message})

        chat_completion = client.chat.completions.create(
            model=model_name, messages=conversation)

        response_message = chat_completion.choices[0].message
        output = response_message.content

        conversation.append(response_message)  # type: ignore
        print(output)

subparser_init 

subparser_init(
    subparsers: _SubParsersAction,
) -> FlexibleArgumentParser
Source code in vllm/entrypoints/cli/openai.py 
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
def subparser_init(
        self,
        subparsers: argparse._SubParsersAction) -> FlexibleArgumentParser:
    chat_parser = subparsers.add_parser(
        "chat",
        help="Generate chat completions via the running API server.",
        description="Generate chat completions via the running API server.",
        usage="vllm chat [options]")
    _add_query_options(chat_parser)
    chat_parser.add_argument(
        "--system-prompt",
        type=str,
        default=None,
        help=("The system prompt to be added to the chat template, "
              "used for models that support system prompts."))
    chat_parser.add_argument("-q",
                             "--quick",
                             type=str,
                             metavar="MESSAGE",
                             help=("Send a single prompt as MESSAGE "
                                   "and print the response, then exit."))
    return chat_parser

CompleteCommand

Bases: CLISubcommand

The complete subcommand for the vLLM CLI.

Source code in vllm/entrypoints/cli/openai.py 
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
class CompleteCommand(CLISubcommand):
    """The `complete` subcommand for the vLLM CLI. """
    name = 'complete'

    @staticmethod
    def cmd(args: argparse.Namespace) -> None:
        model_name, client = _interactive_cli(args)

        if args.quick:
            completion = client.completions.create(model=model_name,
                                                   prompt=args.quick)
            print(completion.choices[0].text)
            return

        print("Please enter prompt to complete:")
        while True:
            input_prompt = input("> ")
            completion = client.completions.create(model=model_name,
                                                   prompt=input_prompt)
            output = completion.choices[0].text
            print(output)

    def subparser_init(
            self,
            subparsers: argparse._SubParsersAction) -> FlexibleArgumentParser:
        complete_parser = subparsers.add_parser(
            "complete",
            help=("Generate text completions based on the given prompt "
                  "via the running API server."),
            description=("Generate text completions based on the given prompt "
                         "via the running API server."),
            usage="vllm complete [options]")
        _add_query_options(complete_parser)
        complete_parser.add_argument(
            "-q",
            "--quick",
            type=str,
            metavar="PROMPT",
            help=
            "Send a single prompt and print the completion output, then exit.")
        return complete_parser

name class-attribute instance-attribute 

name = 'complete'

cmd staticmethod 

cmd(args: Namespace) -> None
Source code in vllm/entrypoints/cli/openai.py 
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
@staticmethod
def cmd(args: argparse.Namespace) -> None:
    model_name, client = _interactive_cli(args)

    if args.quick:
        completion = client.completions.create(model=model_name,
                                               prompt=args.quick)
        print(completion.choices[0].text)
        return

    print("Please enter prompt to complete:")
    while True:
        input_prompt = input("> ")
        completion = client.completions.create(model=model_name,
                                               prompt=input_prompt)
        output = completion.choices[0].text
        print(output)

subparser_init 

subparser_init(
    subparsers: _SubParsersAction,
) -> FlexibleArgumentParser
Source code in vllm/entrypoints/cli/openai.py 
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
def subparser_init(
        self,
        subparsers: argparse._SubParsersAction) -> FlexibleArgumentParser:
    complete_parser = subparsers.add_parser(
        "complete",
        help=("Generate text completions based on the given prompt "
              "via the running API server."),
        description=("Generate text completions based on the given prompt "
                     "via the running API server."),
        usage="vllm complete [options]")
    _add_query_options(complete_parser)
    complete_parser.add_argument(
        "-q",
        "--quick",
        type=str,
        metavar="PROMPT",
        help=
        "Send a single prompt and print the completion output, then exit.")
    return complete_parser

_add_query_options 

_add_query_options(
    parser: FlexibleArgumentParser,
) -> FlexibleArgumentParser
Source code in vllm/entrypoints/cli/openai.py 
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
def _add_query_options(
        parser: FlexibleArgumentParser) -> FlexibleArgumentParser:
    parser.add_argument(
        "--url",
        type=str,
        default="http://localhost:8000/v1",
        help="url of the running OpenAI-Compatible RESTful API server")
    parser.add_argument(
        "--model-name",
        type=str,
        default=None,
        help=("The model name used in prompt completion, default to "
              "the first model in list models API call."))
    parser.add_argument(
        "--api-key",
        type=str,
        default=None,
        help=(
            "API key for OpenAI services. If provided, this api key "
            "will overwrite the api key obtained through environment variables."
        ))
    return parser

_interactive_cli 

_interactive_cli(args: Namespace) -> tuple[str, OpenAI]
Source code in vllm/entrypoints/cli/openai.py 
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
def _interactive_cli(args: argparse.Namespace) -> tuple[str, OpenAI]:
    _register_signal_handlers()

    base_url = args.url
    api_key = args.api_key or os.environ.get("OPENAI_API_KEY", "EMPTY")
    openai_client = OpenAI(api_key=api_key, base_url=base_url)

    if args.model_name:
        model_name = args.model_name
    else:
        available_models = openai_client.models.list()
        model_name = available_models.data[0].id

    print(f"Using model: {model_name}")

    return model_name, openai_client

_register_signal_handlers 

_register_signal_handlers()
Source code in vllm/entrypoints/cli/openai.py 
21
22
23
24
25
26
27
def _register_signal_handlers():

    def signal_handler(sig, frame):
        sys.exit(0)

    signal.signal(signal.SIGINT, signal_handler)
    signal.signal(signal.SIGTSTP, signal_handler)

chat 

chat(
    system_prompt: str | None,
    model_name: str,
    client: OpenAI,
) -> None
Source code in vllm/entrypoints/cli/openai.py 
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
def chat(system_prompt: str | None, model_name: str, client: OpenAI) -> None:
    conversation: list[ChatCompletionMessageParam] = []
    if system_prompt is not None:
        conversation.append({"role": "system", "content": system_prompt})

    print("Please enter a message for the chat model:")
    while True:
        try:
            input_message = input("> ")
        except EOFError:
            return
        conversation.append({"role": "user", "content": input_message})

        chat_completion = client.chat.completions.create(model=model_name,
                                                         messages=conversation)

        response_message = chat_completion.choices[0].message
        output = response_message.content

        conversation.append(response_message)  # type: ignore
        print(output)

cmd_init 

cmd_init() -> list[CLISubcommand]
Source code in vllm/entrypoints/cli/openai.py 
200
201
def cmd_init() -> list[CLISubcommand]:
    return [ChatCommand(), CompleteCommand()]