Skip to content

fix deepseek bug in stream mode #4118

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

Closed
wants to merge 2 commits into from
Closed

fix deepseek bug in stream mode #4118

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
506aada
fix deepseek bug in stream mode
qhduan Nov 17, 2023
af5a6ce
remove multibyte_pending from server
qhduan Nov 18, 2023
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
65 changes: 33 additions & 32 deletions examples/server/server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -362,7 +362,6 @@ struct llama_client_slot

int32_t num_prompt_tokens = 0;
int32_t num_prompt_tokens_processed = 0;
int32_t multibyte_pending = 0;

json prompt;
std::string generated_text;
Expand Down Expand Up @@ -405,7 +404,6 @@ struct llama_client_slot
stopped_word = false;
stopped_limit = false;
stopping_word = "";
multibyte_pending = 0;
n_past = 0;
sent_count = 0;
sent_token_probs_index = 0;
Expand Down Expand Up @@ -949,6 +947,37 @@ struct llama_server_context
return stop_pos;
}

bool is_valid_utf8(const std::string& str) {
int bytesToProcess = 0;

for (unsigned char c : str) {
if (bytesToProcess == 0) {
if ((c >> 7) == 0x0) {
// 1-byte character
continue;
} else if ((c >> 5) == 0x6) {
// 2-byte character: 110xxxxx 10xxxxxx
bytesToProcess = 1;
} else if ((c >> 4) == 0xE) {
// 3-byte character: 1110xxxx 10xxxxxx 10xxxxxx
bytesToProcess = 2;
} else if ((c >> 3) == 0x1E) {
// 4-byte character: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
bytesToProcess = 3;
} else {
return false; // Invalid first byte of a character
}
} else {
if ((c >> 6) != 0x2) { // check this: 10xxxxxx
return false; // Invalid subsequent byte
}
--bytesToProcess;
}
}

return bytesToProcess == 0; // True if all characters were processed completely
}

bool process_token(completion_token_output &result, llama_client_slot &slot) {
// remember which tokens were sampled - used for repetition penalties during sampling
const std::string token_str = llama_token_to_piece(ctx, result.tok);
Expand All @@ -958,35 +987,7 @@ struct llama_server_context
slot.generated_text += token_str;
slot.has_next_token = true;

if (slot.multibyte_pending > 0)
{
slot.multibyte_pending -= token_str.size();
}
else if (token_str.size() == 1)
{
const char c = token_str[0];
// 2-byte characters: 110xxxxx 10xxxxxx
if ((c & 0xE0) == 0xC0)
{
slot.multibyte_pending = 1;
// 3-byte characters: 1110xxxx 10xxxxxx 10xxxxxx
}
else if ((c & 0xF0) == 0xE0)
{
slot.multibyte_pending = 2;
// 4-byte characters: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
}
else if ((c & 0xF8) == 0xF0)
{
slot.multibyte_pending = 3;
}
else
{
slot.multibyte_pending = 0;
}
}

if (slot.multibyte_pending == 0)
if (is_valid_utf8(token_str))
{
size_t pos = std::min(slot.sent_count, slot.generated_text.size());
const std::string str_test = slot.generated_text.substr(pos);
Expand Down Expand Up @@ -1021,7 +1022,7 @@ struct llama_server_context
}
}

if (slot.multibyte_pending > 0 && !slot.has_next_token)
if (!is_valid_utf8(token_str) && !slot.has_next_token)
{
slot.has_next_token = true;
}
Expand Down