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

server : fix smart selection of available slot (#10120)

Browse source 
* Fix smart selection of available slot

* minor fix

* replace vectors of tokens with shorthands
This commit is contained in:
sasha0552 2024-11-01 13:33:14 +00:00 • committed by GitHub
parent 1804adb0cf
commit d865d1478c
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 59 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

35
examples/server/server.cpp
View file

@ -725,12 +725,12 @@ struct server_context {
return nullptr;
}
server_slot * get_available_slot(const std::string & prompt) {
server_slot * get_available_slot(const server_task & task) {
server_slot * ret = nullptr;
// find the slot that has at least n% prompt similarity
if (ret == nullptr && slot_prompt_similarity != 0.0f && !prompt.empty()) {
int max_lcp_len = 0;
if (ret == nullptr && slot_prompt_similarity != 0.0f) {
int max_lcs_len = 0;
float similarity = 0;
for (server_slot & slot : slots) {
@ -740,25 +740,25 @@ struct server_context {
}
// skip the slot if it does not contains cached tokens
if (slot.prompt_tokens.empty()) {
if (slot.cache_tokens.empty()) {
continue;
}
// length of the Longest Common Prefix between the current slot's prompt and the input prompt
int lcp_len = longest_common_prefix(slot.cache_tokens, slot.prompt_tokens);
// length of the Longest Common Subsequence between the current slot's prompt and the input prompt
int lcs_len = longest_common_subsequence(slot.cache_tokens, task.prompt_tokens);
// fraction of the common substring length compared to the current slot's prompt length
similarity = static_cast<float>(lcp_len) / static_cast<int>(slot.prompt_tokens.size());
// fraction of the common subsequence length compared to the current slot's prompt length
similarity = static_cast<float>(lcs_len) / static_cast<int>(slot.cache_tokens.size());
// select the current slot if the criteria match
if (lcp_len > max_lcp_len && similarity > slot_prompt_similarity) {
max_lcp_len = lcp_len;
if (lcs_len > max_lcs_len && similarity > slot_prompt_similarity) {
max_lcs_len = lcs_len;
ret = &slot;
}
}
if (ret != nullptr) {
SLT_DBG(*ret, "selected slot by lcp similarity, max_lcp_len = %d, similarity = %f\n", max_lcp_len, similarity);
SLT_DBG(*ret, "selected slot by lcs similarity, max_lcs_len = %d, similarity = %f\n", max_lcs_len, similarity);
}
}
@ -1514,18 +1514,7 @@ struct server_context {
{
const int id_slot = json_value(task.data, "id_slot", -1);
server_slot * slot;
if (id_slot != -1) {
slot = get_slot_by_id(id_slot);
} else {
std::string prompt;
if (task.data.contains("prompt") && task.data.at("prompt").is_string()) {
prompt = json_value(task.data, "prompt", std::string());
}
slot = get_available_slot(prompt);
}
server_slot * slot = id_slot != -1 ? get_slot_by_id(id_slot) : get_available_slot(task);
if (slot == nullptr) {
// if no slot is available, we defer this task for processing later