fix: adds index, model id, system fingerprint and updates do_sample param

router/src/lib.rs

@@ -158,7 +158,7 @@ fn default_parameters() -> GenerateParameters {
         top_k: None,
         top_p: None,
         typical_p: None,
-        do_sample: false,
+        do_sample: true,
         max_new_tokens: default_max_new_tokens(),
         return_full_text: None,
         stop: Vec::new(),
@@ -253,21 +253,29 @@ pub(crate) struct ChatCompletionDelta {
 }
 
 impl ChatCompletionChunk {
-    pub(crate) fn new(delta: String, created: u64, index: u32) -> Self {
+    pub(crate) fn new(
+        model: String,
+        system_fingerprint: String,
+        delta: String,
+        created: u64,
+        index: u32,
+        logprobs: Option<Vec<f32>>,
+        finish_reason: Option<String>,
+    ) -> Self {
         Self {
             id: "".to_string(),
             object: "text_completion".to_string(),
             created,
-            model: "".to_string(),
-            system_fingerprint: "".to_string(),
+            model,
+            system_fingerprint,
             choices: vec![ChatCompletionChoice {
                 index,
                 delta: ChatCompletionDelta {
                     role: "assistant".to_string(),
                     content: delta,
                 },
-                logprobs: None,
-                finish_reason: None,
+                logprobs,
+                finish_reason,
             }],
         }
     }

router/src/server.rs

@@ -21,7 +21,6 @@ use metrics_exporter_prometheus::{Matcher, PrometheusBuilder, PrometheusHandle};
 use std::convert::Infallible;
 use std::net::SocketAddr;
 use std::sync::atomic::AtomicBool;
-use std::sync::atomic::{AtomicU32, Ordering};
 use std::sync::Arc;
 use text_generation_client::{ShardInfo, ShardedClient};
 use tokenizers::Tokenizer;
@@ -339,7 +338,7 @@ async fn generate_stream(
     HeaderMap,
     Sse<impl Stream<Item = Result<Event, Infallible>>>,
 ) {
-    let on_message_callback = |stream_token: StreamResponse| {
+    let on_message_callback = |_: u32, stream_token: StreamResponse| {
         let event = Event::default();
         event.json_data(stream_token).unwrap()
     };
@@ -353,7 +352,7 @@ async fn generate_stream(
 async fn generate_stream_internal(
     infer: Infer,
     Json(req): Json<GenerateRequest>,
-    on_message_callback: impl Fn(StreamResponse) -> Event,
+    on_message_callback: impl Fn(u32, StreamResponse) -> Event,
 ) -> (HeaderMap, impl Stream<Item = Result<Event, Infallible>>) {
     let span = tracing::Span::current();
     let start_time = Instant::now();
@@ -397,8 +396,10 @@ async fn generate_stream_internal(
             match infer.generate_stream(req).instrument(info_span!(parent: &span, "async_stream")).await {
                 // Keep permit as long as generate_stream lives
                 Ok((_permit, mut response_stream)) => {
+                    let mut index = 0;
                     // Server-Sent Event stream
                     while let Some(response) = response_stream.next().await {
+                        index += 1;
                         match response {
                             Ok(response) => {
                                 match response {
@@ -418,8 +419,7 @@ async fn generate_stream_internal(
                                             generated_text: None,
                                             details: None,
                                         };
-
-                                        let event = on_message_callback(stream_token);
+                                        let event = on_message_callback(index, stream_token);
                                         yield Ok(event);
                                     }
                                     // Yield event for last token and compute timings
@@ -483,7 +483,7 @@ async fn generate_stream_internal(
                                         };
 
 
-                                        let event = on_message_callback(stream_token);
+                                        let event = on_message_callback(index, stream_token);
                                         yield Ok(event);
                                         break;
                                     }
@@ -550,6 +550,7 @@ async fn generate_stream_internal(
     )]
 async fn chat_completions(
     Extension(infer): Extension<Infer>,
+    Extension(info): Extension<Info>,
     Json(req): Json<ChatRequest>,
 ) -> Result<Response, (StatusCode, Json<ErrorResponse>)> {
     metrics::increment_counter!("tgi_request_count");
@@ -605,9 +606,14 @@ async fn chat_completions(
 
     // switch on stream
     if stream {
-        let stream_count = AtomicU32::new(0);
+        let model_id = info.model_id.clone();
+        let system_fingerprint = format!(
+            "{}-{}",
+            info.version,
+            info.docker_label.unwrap_or("native")
+        );
         // pass this callback to the stream generation and build the required event structure
-        let on_message_callback = move |stream_token: StreamResponse| {
+        let on_message_callback = move |index: u32, stream_token: StreamResponse| {
             let event = Event::default();
 
             let current_time = std::time::SystemTime::now()
@@ -615,15 +621,15 @@ async fn chat_completions(
                 .unwrap_or_else(|_| std::time::Duration::from_secs(0))
                 .as_secs();
 
-            // increment the stream count
-            stream_count.fetch_add(1, Ordering::SeqCst);
-            let current_stream_count = stream_count.load(Ordering::SeqCst);
-
             event
                 .json_data(ChatCompletionChunk::new(
+                    model_id.clone(),
+                    system_fingerprint.clone(),
                     stream_token.token.text,
                     current_time,
-                    current_stream_count,
+                    index,
+                    None,
+                    None,
                 ))
                 .unwrap_or_else(|_| {
                     println!("Failed to serialize ChatCompletionChunk");