A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 10, 10:15 UTC
Rename --exclude-log-deltas to --enable-log-deltas (#32020)
Signed-off-by: Catacomba <kevinsuc16@gmail.com>
Kevin Šuc committed
Jan 9, 2026 at 16:30 UTC
ac9f9330e662261242f31f4e0efc9cb9603e4eae
3 files changed
+9
-10
vllm/entrypoints/openai/api_server.py
+1
-1
@@ -1091,7 +1091,7 @@ async def init_app_state(
1091
enable_prompt_tokens_details=args.enable_prompt_tokens_details,
1092
enable_force_include_usage=args.enable_force_include_usage,
1093
enable_log_outputs=args.enable_log_outputs,
1094
- exclude_log_deltas=args.exclude_log_deltas,
1094
+ enable_log_deltas=args.enable_log_deltas,
1095
log_error_stack=args.log_error_stack,
1096
)
1097
if "generate" in supported_tasks
vllm/entrypoints/openai/cli_args.py
+5
-6
@@ -185,11 +185,12 @@ class FrontendArgs:
185
"""Enable the `/tokenizer_info` endpoint. May expose chat
186
templates and other tokenizer configuration."""
187
enable_log_outputs: bool = False
188
- """If True, log model outputs (generations).
188
+ """If set to True, log model outputs (generations).
189
Requires --enable-log-requests."""
190
- exclude_log_deltas: bool = False
191
- """If True, model outputs will be logged once streaming is complete. Deltas
192
- will not be logged. Requires --enable-log-outputs."""
190
+ enable_log_deltas: bool = True
191
+ """If set to False, output deltas will not be logged. Relevant only if
192
+ --enable-log-outputs is set.
193
+ """
194
h11_max_incomplete_event_size: int = H11_MAX_INCOMPLETE_EVENT_SIZE_DEFAULT
195
"""Maximum size (bytes) of an incomplete HTTP event (header or body) for
196
h11 parser. Helps mitigate header abuse. Default: 4194304 (4 MB)."""
@@ -308,8 +309,6 @@ def validate_parsed_serve_args(args: argparse.Namespace):
309
# Enable auto tool needs a tool call parser to be valid
310
if args.enable_auto_tool_choice and not args.tool_call_parser:
311
raise TypeError("Error: --enable-auto-tool-choice requires --tool-call-parser")
311
- if args.exclude_log_deltas and not args.enable_log_outputs:
312
- raise TypeError("Error: --exclude-log-deltas requires --enable-log-outputs")
312
if args.enable_log_outputs and not args.enable_log_requests:
313
raise TypeError("Error: --enable-log-outputs requires --enable-log-requests")
314
vllm/entrypoints/openai/serving_chat.py
+3
-3
@@ -103,7 +103,7 @@ class OpenAIServingChat(OpenAIServing):
103
enable_prompt_tokens_details: bool = False,
104
enable_force_include_usage: bool = False,
105
enable_log_outputs: bool = False,
106
- exclude_log_deltas: bool = False,
106
+ enable_log_deltas: bool = True,
107
log_error_stack: bool = False,
108
default_chat_template_kwargs: dict[str, Any] | None = None,
109
) -> None:
@@ -121,7 +121,7 @@ class OpenAIServingChat(OpenAIServing):
121
self.trust_request_chat_template = trust_request_chat_template
122
self.default_chat_template_kwargs = default_chat_template_kwargs or {}
123
self.enable_log_outputs = enable_log_outputs
124
- self.exclude_log_deltas = exclude_log_deltas
124
+ self.enable_log_deltas = enable_log_deltas
125
126
# set up logits processors
127
self.logits_processors = self.model_config.logits_processors
@@ -1143,7 +1143,7 @@ class OpenAIServingChat(OpenAIServing):
1143
if tc.function and tc.function.arguments
1144
)
1145
1146
- if delta_content and not self.exclude_log_deltas:
1146
+ if delta_content and self.enable_log_deltas:
1147
self.request_logger.log_outputs(
1148
request_id=request_id,
1149
outputs=delta_content,