CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix llama-run n_past

Browse Source
This commit is contained in:
Xuan Son Nguyen
2025-03-14 10:32:43 +01:00
parent bfdddbc150
commit 5e6a6d4e1c
1 changed files with 5 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
examples/run/run.cpp
View File

@@ -595,6 +595,7 @@ class LlamaData {
std::vector<llama_chat_message> messages; // TODO: switch to common_chat_msg std::vector<llama_chat_message> messages; // TODO: switch to common_chat_msg
std::list<std::string> msg_strs; std::list<std::string> msg_strs;
std::vector<char> fmtted; std::vector<char> fmtted;
llama_pos n_past = 0;
int init(Opt & opt) { int init(Opt & opt) {
model = initialize_model(opt); model = initialize_model(opt);
@@ -946,7 +947,7 @@ static int generate(LlamaData & llama_data, const std::string & prompt, std::str
} }
// prepare a batch for the prompt // prepare a batch for the prompt
llama_batch_ext_ptr batch(llama_batch_ext_init_from_text(tokens.data(), tokens.size(), 0, 0, true)); llama_batch_ext_ptr batch(llama_batch_ext_init_from_text(tokens.data(), tokens.size(), llama_data.n_past, 0, true));
llama_token new_token_id; llama_token new_token_id;
while (true) { while (true) {
check_context_size(llama_data.context, batch); check_context_size(llama_data.context, batch);
@@ -955,6 +956,8 @@ static int generate(LlamaData & llama_data, const std::string & prompt, std::str
return 1; return 1;
} }
llama_data.n_past += llama_batch_ext_get_n_tokens(batch.get());
// sample the next token, check is it an end of generation? // sample the next token, check is it an end of generation?
new_token_id = llama_sampler_sample(llama_data.sampler.get(), llama_data.context.get(), -1); new_token_id = llama_sampler_sample(llama_data.sampler.get(), llama_data.context.get(), -1);
if (llama_vocab_is_eog(vocab, new_token_id)) { if (llama_vocab_is_eog(vocab, new_token_id)) {
@@ -969,7 +972,7 @@ static int generate(LlamaData & llama_data, const std::string & prompt, std::str
print_word_and_concatenate_to_response(piece, response); print_word_and_concatenate_to_response(piece, response);
// prepare the next batch with the sampled token // prepare the next batch with the sampled token
batch.reset(llama_batch_ext_init_from_text(&new_token_id, 1, 0, 0, true)); batch.reset(llama_batch_ext_init_from_text(&new_token_id, 1, llama_data.n_past, 0, true));
} }
printf(LOG_COL_DEFAULT); printf(LOG_COL_DEFAULT);