Return prompt vs generated tokens. (#1436)

# What does this PR do?

Fixes #637 
 
<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Did you read the [contributor
guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? Please add a link
      to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes?
Here are the
[documentation
guidelines](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?


## Who can review?

Anyone in the community is free to review the PR once the tests have
passed. Feel free to tag
members/contributors who may be interested in your PR.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->

router/src/infer.rs

@@ -90,6 +90,7 @@ impl Infer {
     ) -> Result<
         (
             OwnedSemaphorePermit,
+            u32,
             UnboundedReceiverStream<Result<InferStreamResponse, InferError>>,
         ),
         InferError,
@@ -114,6 +115,7 @@ impl Infer {
 
         // MPSC channel to communicate with the background batching task
         let (response_tx, response_rx) = mpsc::unbounded_channel();
+        let input_length = valid_request.input_length;
 
         // Append the request to the queue
         self.queue.append(Entry {
@@ -130,7 +132,11 @@ impl Infer {
         self.shared.batching_task.notify_one();
 
         // Return stream
-        Ok((permit, UnboundedReceiverStream::new(response_rx)))
+        Ok((
+            permit,
+            input_length,
+            UnboundedReceiverStream::new(response_rx),
+        ))
     }
 
     /// Add a new request to the queue and return a InferResponse
@@ -142,7 +148,7 @@ impl Infer {
         let use_top_tokens = request.parameters.top_n_tokens.is_some_and(|x| x > 0);
 
         // Create stream and keep semaphore permit as long as generate lives
-        let (_permit, mut stream) = self.generate_stream(request).await?;
+        let (_permit, _input_length, mut stream) = self.generate_stream(request).await?;
 
         // Return values
         let mut result_prefill = Vec::new();
@@ -196,6 +202,7 @@ impl Infer {
         {
             Ok(InferResponse {
                 prefill: result_prefill,
+                _input_length,
                 tokens: result_tokens,
                 generated_text,
                 queued,
@@ -636,6 +643,10 @@ pub(crate) enum InferStreamResponse {
 
 #[derive(Debug)]
 pub(crate) struct InferResponse {
+    /// input_length is the input as perceived by the rust tokenizer in the
+    /// validation pathway. It is redundant with prefill.len() but prefill
+    /// has data only if the user asked for it. This will always be filled.
+    pub(crate) _input_length: u32,
     pub(crate) prefill: Vec<PrefillToken>,
     pub(crate) tokens: Vec<Token>,
     pub(crate) generated_text: GeneratedText,

router/src/server.rs

@@ -170,6 +170,7 @@ async fn generate(
     };
 
     // Token details
+    let input_length = response._input_length;
     let details = match details {
         true => {
             // convert best_of_responses
@@ -257,6 +258,11 @@ async fn generate(
         "x-time-per-token",
         time_per_token.as_millis().to_string().parse().unwrap(),
     );
+    headers.insert("x-prompt-tokens", input_length.into());
+    headers.insert(
+        "x-generated-tokens",
+        response.generated_text.generated_tokens.into(),
+    );
 
     // Metrics
     metrics::increment_counter!("tgi_request_success");
@@ -378,7 +384,7 @@ async fn generate_stream(
         } else {
             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)) => {
+                Ok((_permit, _input_length, mut response_stream)) => {
                     // Server-Sent Event stream
                     while let Some(response) = response_stream.next().await {
                         match response {