huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 04:14:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

change ChatCompletionChunk to align with "OpenAI Chat Completions streaming API"

Browse Source 
When stream_options: {"include_usage": true} is included, choices is None only for the last chunk, and usage is always None except for the last chunk.

Signed-off-by: Wang, Yi A <yi.a.wang@intel.com>
This commit is contained in:
Wang, Yi A 2025-02-06 21:11:53 -08:00
parent 4b8cda684b
commit efeef0bed6
2 changed files with 64 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
router/src/lib.rs
View File

@ -790,12 +790,15 @@ impl ChatCompletionChunk {
created,
model,
system_fingerprint,
choices: vec![ChatCompletionChoice {
choices: match usage {
None => vec![ChatCompletionChoice {
index: 0,
delta,
logprobs,
finish_reason,
}],
_ => vec![],
},
usage,
}
}

66
router/src/server.rs
View File

@ -1124,7 +1124,6 @@ enum StreamState {
fn create_event_from_stream_token(
stream_token: &StreamResponse,
logprobs: bool,
stream_options: Option<StreamOptions>,
inner_using_tools: bool,
system_fingerprint: String,
model_id: String,
@ -1153,8 +1152,43 @@ fn create_event_from_stream_token(
};
let (usage, finish_reason) = match &stream_token.details {
Some(details) => (None, Some(details.finish_reason.format(true))),
None => (None, None),
};
let chat_complete = CompletionType::ChatCompletionChunk(ChatCompletionChunk::new(
model_id.clone(),
system_fingerprint.clone(),
content,
tool_calls,
current_time,
logprobs,
finish_reason,
usage,
));
event.json_data(chat_complete).unwrap_or_else(|e| {
println!("Failed to serialize ChatCompletionChunk: {:?}", e);
Event::default()
})
}
/// Convert a StreamResponse into an Event to be sent over SSE
fn create_usage_event_from_stream_token(
stream_token: &StreamResponse,
stream_options: Option<StreamOptions>,
system_fingerprint: String,
model_id: String,
) -> Event {
let event = Event::default();
let current_time = std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.unwrap_or_else(|_| std::time::Duration::from_secs(0))
.as_secs();
let usage = match &stream_token.details {
Some(details) => {
let usage = if stream_options
if stream_options
.as_ref()
.map(|s| s.include_usage)
.unwrap_or(false)
@ -1169,20 +1203,19 @@ fn create_event_from_stream_token(
})
} else {
None
};
(usage, Some(details.finish_reason.format(true)))
}
None => (None, None),
}
None => None,
};
let chat_complete = CompletionType::ChatCompletionChunk(ChatCompletionChunk::new(
model_id.clone(),
system_fingerprint.clone(),
content,
tool_calls,
None,
None,
current_time,
logprobs,
finish_reason,
None,
None,
usage,
));
@ -1307,7 +1340,6 @@ pub(crate) async fn chat_completions(
let event = create_event_from_stream_token(
stream_token,
logprobs,
stream_options.clone(),
response_as_tool,
system_fingerprint.clone(),
model_id.clone(),
@ -1369,13 +1401,25 @@ pub(crate) async fn chat_completions(
let event = create_event_from_stream_token(
&stream_token,
logprobs,
stream_options.clone(),
response_as_tool,
system_fingerprint.clone(),
model_id.clone(),
);
yield Ok::<Event, Infallible>(event);
if stream_token.details.is_some() && stream_options
.as_ref()
.map(|s| s.include_usage)
.unwrap_or(false) {
let usage_event = create_usage_event_from_stream_token(
&stream_token,
stream_options.clone(),
system_fingerprint.clone(),
model_id.clone(),
);
yield Ok::<Event, Infallible>(usage_event);
}
}
}
}