ollama/llm/llama.cpp/ggml_patch/0001-add-detokenize-endpoint.patch
Bruce MacDonald 42998d797d
subprocess llama.cpp server (#401)
* remove c code
* pack llama.cpp
* use request context for llama_cpp
* let llama_cpp decide the number of threads to use
* stop llama runner when app stops
* remove sample count and duration metrics
* use go generate to get libraries
* tmp dir for running llm
2023-08-30 16:35:03 -04:00

51 lines
1.7 KiB
Diff

From 032ef7ff2423f5117bb59d42fb71be9cebf0a2de Mon Sep 17 00:00:00 2001
From: Bruce MacDonald <brucewmacdonald@gmail.com>
Date: Mon, 28 Aug 2023 18:08:12 -0400
Subject: [PATCH] add detokenize endpoint
---
examples/server/server.cpp | 21 +++++++++++++++++++++
1 file changed, 21 insertions(+)
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index 9966045..5014691 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -1075,6 +1075,12 @@ static json format_tokenizer_response(const std::vector<llama_token> &tokens)
{"tokens", tokens}};
}
+static json format_detokenized_response(std::string content)
+{
+ return json{
+ {"content", content}};
+}
+
static void parse_options_completion(const json &body, llama_server_context &llama)
{
gpt_params default_params;
@@ -1361,6 +1367,21 @@ int main(int argc, char **argv)
const json data = format_tokenizer_response(tokens);
return res.set_content(data.dump(), "application/json"); });
+ svr.Post("/detokenize", [&llama](const Request &req, Response &res)
+ {
+ auto lock = llama.lock();
+
+ const json body = json::parse(req.body);
+ std::string content;
+ if (body.count("tokens") != 0)
+ {
+ const std::vector<llama_token> tokens = body["tokens"];
+ content = tokens_to_str(llama.ctx, tokens.cbegin(), tokens.cend());
+ }
+
+ const json data = format_detokenized_response(content);
+ return res.set_content(data.dump(), "application/json"); });
+
svr.Post("/embedding", [&llama](const Request &req, Response &res)
{
auto lock = llama.lock();
--
2.39.2 (Apple Git-143)