Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

server: allow filtering llama server response fields #10940

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 5 additions & 1 deletion examples/server/server.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -91,6 +91,7 @@ struct slot_params {
int64_t t_max_predict_ms = -1; // if positive, limit the generation phase to this time limit

std::vector<std::string> antiprompt;
std::vector<std::string> requested_fields;
bool timings_per_token = false;
bool ignore_eos = false;

Expand Down Expand Up @@ -205,6 +206,7 @@ struct server_task {
params.n_discard = json_value(data, "n_discard", defaults.n_discard);
//params.t_max_prompt_ms = json_value(data, "t_max_prompt_ms", defaults.t_max_prompt_ms); // TODO: implement
params.t_max_predict_ms = json_value(data, "t_max_predict_ms", defaults.t_max_predict_ms);
params.requested_fields = json_value(data, "requested_fields", std::vector<std::string>());

params.sampling.top_k = json_value(data, "top_k", defaults.sampling.top_k);
params.sampling.top_p = json_value(data, "top_p", defaults.sampling.top_p);
Expand Down Expand Up @@ -482,6 +484,7 @@ struct server_task_result_cmpl_final : server_task_result {
stop_type stop = STOP_TYPE_NONE;

std::vector<completion_token_output> probs_output;
std::vector<std::string> requested_fields;

slot_params generation_params;

Expand Down Expand Up @@ -527,7 +530,7 @@ struct server_task_result_cmpl_final : server_task_result {
if (!probs_output.empty()) {
res["completion_probabilities"] = completion_token_output::probs_vector_to_json(probs_output);
}
return res;
return requested_fields.empty() ? res : json_get_nested_values(requested_fields, res);
}

json to_json_oaicompat_chat() {
Expand Down Expand Up @@ -1960,6 +1963,7 @@ struct server_context {
res->content = slot.generated_text;
res->timings = slot.get_timings();
res->prompt = common_detokenize(ctx, slot.prompt_tokens, true);
res->requested_fields = slot.params.requested_fields;

res->truncated = slot.truncated;
res->n_decoded = slot.n_decoded;
Expand Down
27 changes: 27 additions & 0 deletions examples/server/utils.hpp
Original file line number Diff line number Diff line change
Expand Up @@ -88,6 +88,33 @@ static bool json_is_array_of_mixed_numbers_strings(const json & data) {
return false;
}

// get value by path(key1 / key2)
static json json_get_nested_values(const std::vector<std::string>& paths, const json& js) {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
static json json_get_nested_values(const std::vector<std::string>& paths, const json& js) {
static json json_get_nested_values(const std::vector<std::string> & paths, const json & js) {

You should modify other places in your code to follow this style, as stated in https://github.com/ggerganov/llama.cpp/blob/master/CONTRIBUTING.md

Clean-up any trailing whitespaces, use 4 spaces for indentation, brackets on the same line, void * ptr, int & a

json result = json::object();

for (const std::string& path : paths) {
json current = js;
std::istringstream stream(path);
std::string key;
std::vector<std::string> keys;
while (std::getline(stream, key, '/')) {
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can use string_split in common.h

keys.push_back(key);
}
bool valid_path = true;
for (const std::string& k : keys) {
if (valid_path && current.is_object() && current.contains(k)) {
current = current[k];
} else {
valid_path = false;
}
}
if (valid_path) {
result[path] = current;
}
}
return result;
}

/**
* this handles 2 cases:
* - only string, example: "string"
Expand Down
Loading