Skip to content

vllm.entrypoints.logger

logger module-attribute 

logger = init_logger(__name__)

RequestLogger

Source code in vllm/entrypoints/logger.py 
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
class RequestLogger:
    def __init__(self, *, max_log_len: int | None) -> None:
        self.max_log_len = max_log_len

    def log_inputs(
        self,
        request_id: str,
        prompt: str | None,
        prompt_token_ids: list[int] | None,
        prompt_embeds: torch.Tensor | None,
        params: SamplingParams | PoolingParams | BeamSearchParams | None,
        lora_request: LoRARequest | None,
    ) -> None:
        max_log_len = self.max_log_len
        if max_log_len is not None:
            if prompt is not None:
                prompt = prompt[:max_log_len]

            if prompt_token_ids is not None:
                prompt_token_ids = prompt_token_ids[:max_log_len]

        logger.debug(
            "Request %s details: prompt: %r, "
            "prompt_token_ids: %s, "
            "prompt_embeds shape: %s.",
            request_id,
            prompt,
            prompt_token_ids,
            prompt_embeds.shape if prompt_embeds is not None else None,
        )

        logger.info(
            "Received request %s: params: %s, lora_request: %s.",
            request_id,
            params,
            lora_request,
        )

    def log_outputs(
        self,
        request_id: str,
        outputs: str,
        output_token_ids: Sequence[int] | None,
        finish_reason: str | None = None,
        is_streaming: bool = False,
        delta: bool = False,
    ) -> None:
        max_log_len = self.max_log_len
        if max_log_len is not None:
            if outputs is not None:
                outputs = outputs[:max_log_len]

            if output_token_ids is not None:
                # Convert to list and apply truncation
                output_token_ids = list(output_token_ids)[:max_log_len]

        stream_info = ""
        if is_streaming:
            stream_info = " (streaming delta)" if delta else " (streaming complete)"

        logger.info(
            "Generated response %s%s: output: %r, "
            "output_token_ids: %s, finish_reason: %s",
            request_id,
            stream_info,
            outputs,
            output_token_ids,
            finish_reason,
        )

max_log_len instance-attribute 

max_log_len = max_log_len

__init__ 

__init__(*, max_log_len: int | None) -> None
Source code in vllm/entrypoints/logger.py 
17
18
def __init__(self, *, max_log_len: int | None) -> None:
    self.max_log_len = max_log_len

log_inputs 

log_inputs(
    request_id: str,
    prompt: str | None,
    prompt_token_ids: list[int] | None,
    prompt_embeds: Tensor | None,
    params: SamplingParams
    | PoolingParams
    | BeamSearchParams
    | None,
    lora_request: LoRARequest | None,
) -> None
Source code in vllm/entrypoints/logger.py 
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
def log_inputs(
    self,
    request_id: str,
    prompt: str | None,
    prompt_token_ids: list[int] | None,
    prompt_embeds: torch.Tensor | None,
    params: SamplingParams | PoolingParams | BeamSearchParams | None,
    lora_request: LoRARequest | None,
) -> None:
    max_log_len = self.max_log_len
    if max_log_len is not None:
        if prompt is not None:
            prompt = prompt[:max_log_len]

        if prompt_token_ids is not None:
            prompt_token_ids = prompt_token_ids[:max_log_len]

    logger.debug(
        "Request %s details: prompt: %r, "
        "prompt_token_ids: %s, "
        "prompt_embeds shape: %s.",
        request_id,
        prompt,
        prompt_token_ids,
        prompt_embeds.shape if prompt_embeds is not None else None,
    )

    logger.info(
        "Received request %s: params: %s, lora_request: %s.",
        request_id,
        params,
        lora_request,
    )

log_outputs 

log_outputs(
    request_id: str,
    outputs: str,
    output_token_ids: Sequence[int] | None,
    finish_reason: str | None = None,
    is_streaming: bool = False,
    delta: bool = False,
) -> None
Source code in vllm/entrypoints/logger.py 
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
def log_outputs(
    self,
    request_id: str,
    outputs: str,
    output_token_ids: Sequence[int] | None,
    finish_reason: str | None = None,
    is_streaming: bool = False,
    delta: bool = False,
) -> None:
    max_log_len = self.max_log_len
    if max_log_len is not None:
        if outputs is not None:
            outputs = outputs[:max_log_len]

        if output_token_ids is not None:
            # Convert to list and apply truncation
            output_token_ids = list(output_token_ids)[:max_log_len]

    stream_info = ""
    if is_streaming:
        stream_info = " (streaming delta)" if delta else " (streaming complete)"

    logger.info(
        "Generated response %s%s: output: %r, "
        "output_token_ids: %s, finish_reason: %s",
        request_id,
        stream_info,
        outputs,
        output_token_ids,
        finish_reason,
    )