Update MPI code to new KV seq rm and bos/eos model APIs
This commit is contained in:
parent
bcfb190c28
commit
888d4f591b
1 changed files with 5 additions and 5 deletions
|
@ -284,7 +284,7 @@ int main(int argc, char ** argv) {
|
|||
|
||||
// Should not run without any tokens
|
||||
if (embd_inp.empty()) {
|
||||
embd_inp.push_back(llama_token_bos(ctx));
|
||||
embd_inp.push_back(llama_token_bos(model));
|
||||
LOG("embd_inp was considered empty and bos was added: %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, embd_inp).c_str());
|
||||
}
|
||||
|
||||
|
@ -334,7 +334,7 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
|
||||
// remove any "future" tokens that we might have inherited from the previous session
|
||||
llama_kv_cache_tokens_rm(ctx, n_matching_session_tokens, -1);
|
||||
llama_kv_cache_seq_rm(ctx, -1, n_matching_session_tokens, -1);
|
||||
}
|
||||
|
||||
LOGLN(
|
||||
|
@ -729,7 +729,7 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
|
||||
// deal with end of text token in interactive mode
|
||||
if (llama_sampling_last(ctx_sampling) == llama_token_eos(ctx)) {
|
||||
if (llama_sampling_last(ctx_sampling) == llama_token_eos(model)) {
|
||||
LOG("found EOS token\n");
|
||||
|
||||
if (params.interactive) {
|
||||
|
@ -756,7 +756,7 @@ int main(int argc, char ** argv) {
|
|||
|
||||
if (params.input_prefix_bos) {
|
||||
LOG("adding input prefix BOS token\n");
|
||||
embd_inp.push_back(llama_token_bos(ctx));
|
||||
embd_inp.push_back(llama_token_bos(model));
|
||||
}
|
||||
|
||||
std::string buffer;
|
||||
|
@ -840,7 +840,7 @@ int main(int argc, char ** argv) {
|
|||
}
|
||||
|
||||
// end of text token
|
||||
if (!embd.empty() && embd.back() == llama_token_eos(ctx) && !(params.instruct || params.interactive)) {
|
||||
if (!embd.empty() && embd.back() == llama_token_eos(model) && !(params.instruct || params.interactive)) {
|
||||
LOG_TEE(" [end of text]\n");
|
||||
break;
|
||||
}
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue