Skip to content

server : add n_discard parameter to specify the number of tokens to discard when context is shifted #6300

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.

Sign up for GitHub

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

Merged
merged 1 commit into from
Mar 26, 2024
Merged

server : add n_discard parameter to specify the number of tokens to discard when context is shifted #6300

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
5 changes: 4 additions & 1 deletion examples/server/server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,6 +99,7 @@ struct slot_params {

uint32_t seed = -1; // RNG seed
int32_t n_keep = 0; // number of tokens to keep from initial prompt
int32_t n_discard = 0; // number of tokens after n_keep that may be discarded when shifting context, 0 defaults to half
int32_t n_predict = -1; // new tokens to predict

std::vector<std::string> antiprompt;
Expand Down Expand Up @@ -846,6 +847,7 @@ struct server_context {
slot.sparams.mirostat_eta = json_value(data, "mirostat_eta", default_sparams.mirostat_eta);
slot.sparams.penalize_nl = json_value(data, "penalize_nl", default_sparams.penalize_nl);
slot.params.n_keep = json_value(data, "n_keep", slot.params.n_keep);
slot.params.n_discard = json_value(data, "n_discard", default_params.n_discard);
slot.params.seed = json_value(data, "seed", default_params.seed);
slot.sparams.n_probs = json_value(data, "n_probs", default_sparams.n_probs);
slot.sparams.min_keep = json_value(data, "min_keep", default_sparams.min_keep);
Expand Down Expand Up @@ -1253,6 +1255,7 @@ struct server_context {
{"stop", slot.params.antiprompt},
{"n_predict", slot.params.n_predict}, // TODO: fix duplicate key n_predict
{"n_keep", slot.params.n_keep},
{"n_discard", slot.params.n_discard},
{"ignore_eos", ignore_eos},
{"stream", slot.params.stream},
{"logit_bias", slot.sparams.logit_bias},
Expand Down Expand Up @@ -1696,7 +1699,7 @@ struct server_context {
// Shift context
const int n_keep = slot.params.n_keep + add_bos_token;
const int n_left = (int) system_tokens.size() + slot.n_past - n_keep;
const int n_discard = n_left / 2;
const int n_discard = slot.params.n_discard ? slot.params.n_discard : (n_left / 2);

LOG_INFO("slot context shift", {
{"id_slot", slot.id},
Expand Down