vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

server : clarify /slots endpoint, add is_processing (#10162)

Browse source 
* server : clarify /slots endpoint, add is_processing

* fix tests
This commit is contained in:
Xuan Son Nguyen 2024-11-04 16:33:29 +01:00 • committed by GitHub
parent 6a066b9978
commit 9e0ecfb697
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 18 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

16
examples/server/server.cpp
View file

@ -1566,11 +1566,11 @@ struct server_context {
for (server_slot & slot : slots) {
json slot_data = get_formated_generation(slot);
slot_data["id"] = slot.id;
slot_data["id_task"] = slot.id_task;
slot_data["state"] = slot.state;
slot_data["prompt"] = common_detokenize(ctx, slot.prompt_tokens);
slot_data["next_token"] = {
slot_data["id"] = slot.id;
slot_data["id_task"] = slot.id_task;
slot_data["is_processing"] = slot.is_processing();
slot_data["prompt"] = common_detokenize(ctx, slot.prompt_tokens);
slot_data["next_token"] = {
{"has_next_token", slot.has_next_token},
{"has_new_line", slot.has_new_line},
{"n_remain", slot.n_remaining},
@ -1581,10 +1581,10 @@ struct server_context {
{"stopping_word", slot.stopping_word},
};
if (slot_data["state"] == SLOT_STATE_IDLE) {
n_idle_slots++;
} else {
if (slot.is_processing()) {
n_processing_slots++;
} else {
n_idle_slots++;
}
slots_data.push_back(slot_data);