Merge remote-tracking branch 'origin/master' into tool-call

Author: Olivier Chafik

README.md

@@ -423,7 +423,7 @@ To learn more about model quantization, [read this documentation](examples/quant
 
     </details>
 
-[^1]: [examples/perplexity/README.md](examples/perplexity/README.md)
+[^1]: [examples/perplexity/README.md](./examples/perplexity/README.md)
 [^2]: [https://huggingface.co/docs/transformers/perplexity](https://huggingface.co/docs/transformers/perplexity)
 
 ## [`llama-bench`](examples/llama-bench)

examples/server/README.md

@@ -460,7 +460,7 @@ These words will not be included in the completion, so make sure to add them to
 - Note: In streaming mode (`stream`), only `content`, `tokens` and `stop` will be returned until end of completion. Responses are sent using the [Server-sent events](https://html.spec.whatwg.org/multipage/server-sent-events.html) standard. Note: the browser's `EventSource` interface cannot be used due to its lack of `POST` request support.
 
 - `completion_probabilities`: An array of token probabilities for each completion. The array's length is `n_predict`. Each item in the array has a nested array `top_logprobs`. It contains at **maximum** `n_probs` elements:
-  ```json
+  ```
   {
     "content": "<the generated completion text>",
     "tokens": [ generated token ids if requested ],
@@ -561,7 +561,7 @@ If `with_pieces` is `true`:
 ```
 
 With input 'á' (utf8 hex: C3 A1) on tinyllama/stories260k
-```json
+```
 {
   "tokens": [
     {"id": 198, "piece": [195]}, // hex C3
@@ -576,6 +576,18 @@ With input 'á' (utf8 hex: C3 A1) on tinyllama/stories260k
 
 `tokens`: Set the tokens to detokenize.
 
+### POST `/apply-template`: Apply chat template to a conversation
+
+Uses the server's prompt template formatting functionality to convert chat messages to a single string expected by a chat model as input, but does not perform inference. Instead, the prompt string is returned in the `prompt` field of the JSON response. The prompt can then be modified as desired (for example, to insert "Sure!" at the beginning of the model's response) before sending to `/completion` to generate the chat response.
+
+*Options:*
+
+`messages`: (Required) Chat turns in the same format as `/v1/chat/completions`.
+
+**Response format**
+
+Returns a JSON object with a field `prompt` containing a string of the input messages formatted according to the model's chat template format.
+
 ### POST `/embedding`: Generate embedding of a given text
 
 > [!IMPORTANT]
@@ -768,7 +780,7 @@ Same as the `/v1/embeddings` endpoint.
 
 **Response format**
 
-```json
+```
 [
   {
     "index": 0,

examples/server/server.cpp

@@ -4185,6 +4185,14 @@ int main(int argc, char ** argv) {
         res_ok(res, root);
     };
 
+    const auto handle_apply_template = [&ctx_server, &params, &res_ok](const httplib::Request & req, httplib::Response & res) {
+        auto body = json::parse(req.body);
+        const auto & chat_template = body.contains("tools") && ctx_server.chat_templates.template_tool_use ? *ctx_server.chat_templates.template_tool_use : *ctx_server.chat_templates.template_default;
+        json data = oaicompat_completion_params_parse(body, chat_template, params.use_jinja);
+
+        res_ok(res, {{ "prompt", data.at("prompt") }});
+    };
+
     const auto handle_embeddings = [&handle_embeddings_impl](const httplib::Request & req, httplib::Response & res) {
         handle_embeddings_impl(req, res, OAICOMPAT_TYPE_NONE);
     };
@@ -4361,6 +4369,7 @@ int main(int argc, char ** argv) {
     svr->Post("/v1/reranking",        handle_rerank);
     svr->Post("/tokenize",            handle_tokenize);
     svr->Post("/detokenize",          handle_detokenize);
+    svr->Post("/apply-template",      handle_apply_template);
     // LoRA adapters hotswap
     svr->Get ("/lora-adapters",       handle_lora_adapters_list);
     svr->Post("/lora-adapters",       handle_lora_adapters_apply);
@@ -4439,11 +4448,13 @@ int main(int argc, char ** argv) {
         ctx_server.chat_templates.template_default->source().c_str(),
         common_chat_format_example(*ctx_server.chat_templates.template_default, ctx_server.params_base.use_jinja).c_str());
 
-    ctx_server.queue_tasks.on_new_task(std::bind(
-                &server_context::process_single_task, &ctx_server, std::placeholders::_1));
+    ctx_server.queue_tasks.on_new_task([&ctx_server](const server_task & task) {
+        ctx_server.process_single_task(task);
+    });
 
-    ctx_server.queue_tasks.on_update_slots(std::bind(
-                &server_context::update_slots, &ctx_server));
+    ctx_server.queue_tasks.on_update_slots([&ctx_server]() {
+        ctx_server.update_slots();
+    });
 
     shutdown_handler = [&](int) {
         ctx_server.queue_tasks.terminate();

examples/server/tests/unit/test_chat_completion.py

@@ -121,6 +121,21 @@ def test_chat_template():
     assert res.body["__verbose"]["prompt"] == "<s> <|start_header_id|>system<|end_header_id|>\n\nBook<|eot_id|><|start_header_id|>user<|end_header_id|>\n\nWhat is the best book<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n"
 
 
+def test_apply_chat_template():
+    global server
+    server.chat_template = "command-r"
+    server.start()
+    res = server.make_request("POST", "/apply-template", data={
+        "messages": [
+            {"role": "system", "content": "You are a test."},
+            {"role": "user", "content":"Hi there"},
+        ]
+    })
+    assert res.status_code == 200
+    assert "prompt" in res.body
+    assert res.body["prompt"] == "<|START_OF_TURN_TOKEN|><|SYSTEM_TOKEN|>You are a test.<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|USER_TOKEN|>Hi there<|END_OF_TURN_TOKEN|><|START_OF_TURN_TOKEN|><|CHATBOT_TOKEN|>"
+
+
 @pytest.mark.parametrize("response_format,n_predicted,re_content", [
     ({"type": "json_object", "schema": {"const": "42"}}, 6, "\"42\""),
     ({"type": "json_object", "schema": {"items": [{"type": "integer"}]}}, 10, "[ -3000 ]"),

src/llama-vocab.cpp

@@ -1692,7 +1692,7 @@ void llama_vocab::impl::load(llama_model_loader & ml, const LLM_KV & kv) {
         GGML_ASSERT(!ids.empty() && "model vocab missing newline token");
         linefeed_id = ids[0];
     } else {
-        const std::vector<int> ids = tokenize("\xC4\x8A", false); // U+010A
+        const std::vector<int> ids = tokenize("\n", false);
 
         //GGML_ASSERT(!ids.empty() && "model vocab missing newline token");
         if (ids.empty()) {