vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 09:04:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix access violation when using banned_phrases (#1154)

Browse source
This commit is contained in:
Maya 2024-10-10 16:46:39 +03:00 • committed by GitHub
parent fe5479f286
commit 5c9650d68e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 12 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

15
gpttype_adapter.cpp
View file

@ -3236,7 +3236,10 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
grammar_accept_token(file_format, n_vocab, grammar, id); grammar_accept_token(file_format, n_vocab, grammar, id);
} }
last_n_tokens.erase(last_n_tokens.begin()); if (!last_n_tokens.empty())
{
last_n_tokens.erase(last_n_tokens.begin());
}
last_n_tokens.push_back(id); last_n_tokens.push_back(id);
current_context_tokens.push_back(id); current_context_tokens.push_back(id);
@ -3416,7 +3419,10 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
int sepsize = llava_sep.size(); int sepsize = llava_sep.size();
while(input_consumed < embd_inp.size() && (embd_inp[input_consumed]==LLAVA_TOKEN_IDENTIFIER_A || embd_inp[input_consumed]==LLAVA_TOKEN_IDENTIFIER_B)) while(input_consumed < embd_inp.size() && (embd_inp[input_consumed]==LLAVA_TOKEN_IDENTIFIER_A || embd_inp[input_consumed]==LLAVA_TOKEN_IDENTIFIER_B))
{ {
last_n_tokens.erase(last_n_tokens.begin()); if (!last_n_tokens.empty())
{
last_n_tokens.erase(last_n_tokens.begin());
}
last_n_tokens.push_back(currtoken); last_n_tokens.push_back(currtoken);
current_context_tokens.push_back(currtoken); current_context_tokens.push_back(currtoken);
++input_consumed; ++input_consumed;
@ -3472,7 +3478,10 @@ generation_outputs gpttype_generate(const generation_inputs inputs)
else else
{ {
embd.push_back(currtoken); embd.push_back(currtoken);
last_n_tokens.erase(last_n_tokens.begin()); if (!last_n_tokens.empty())
{
last_n_tokens.erase(last_n_tokens.begin());
}
last_n_tokens.push_back(currtoken); last_n_tokens.push_back(currtoken);
current_context_tokens.push_back(currtoken); current_context_tokens.push_back(currtoken);
++input_consumed; ++input_consumed;