niansa/libjustlm
1
0
Fork 0
mirror of https://gitlab.com/niansa/libjustlm.git synced 2025-03-06 20:49:17 +01:00
Code Releases Activity

Minor improvements on EOS handling

Browse source
This commit is contained in:
niansa/tuxifan 2023-05-17 10:51:20 +02:00
parent 8e7e310757
commit abbb35c6a9
3 changed files with 19 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
justlm_gptj.hpp
View file

@ -179,17 +179,16 @@ public:
continue;
}
id = gpt_tokenize(state->vocab, "\n")[0];
state->tokens.push_back(id);
} else {
// Add token
state->tokens.push_back(id);
}
// Add token
state->tokens.push_back(id);
// Make sure token limit isn't being hit
LM_COAWAIT window_scroll();
// Get token as string
const auto str = state->vocab.id_to_token[id];
const std::string_view str = state->vocab.id_to_token[id];
// Append string to function result
state->prompt.append(str);
@ -203,7 +202,7 @@ public:
}
// Tick
if (on_tick && !on_tick(str.c_str())) abort = true;
if (on_tick && !on_tick(str.data())) abort = true;
else if (!LM_TASKYIELD) abort = true;
}

4
justlm_llama.hpp
View file

@ -180,7 +180,7 @@ public:
LM_COAWAIT window_scroll();
// Get token as string
const auto str = llama_token_to_str(state->ctx, id);
const std::string_view str = llama_token_to_str(state->ctx, id);
// Append string to function result
state->prompt.append(str);
@ -193,7 +193,7 @@ public:
}
// Tick and yield
if (on_tick && !on_tick(str)) abort = true;
if (on_tick && !on_tick(str.data())) abort = true;
else if (!LM_TASKYIELD) abort = true;
}

22
justlm_mpt.hpp
View file

@ -182,25 +182,28 @@ public:
const auto n_repeat_last = std::min<size_t>(state->tokens.size(), params.n_repeat_last);
auto id = gpt_sample_top_k_top_p(state->model.hparams.n_vocab, state->tokens.data()+state->tokens.size()-n_repeat_last, n_repeat_last, state->logits, params.top_k, params.top_p, params.temp, params.repeat_penalty, state->rng);
if (id == 0 || id == state->im_end) {
if (state->im_end && id == state->im_end) {
if (eos_count++ == params.eos_ignores) {
abort = true;
continue;
}
id = gpt_tokenize(state->vocab, "\n")[0];
} else if (id == 0) {
if (eos_count++ == params.eos_ignores) {
abort = true;
printf("Stopping due to EOS (%d)\n", id);
continue;
}
printf("Retrying after EOS (%d)... %d\n", id, eos_count);
id = gpt_tokenize(state->vocab, "\n")[0];
state->tokens.push_back(id);
} else {
// Add token
state->tokens.push_back(id);
}
// Add token
state->tokens.push_back(id);
// Make sure token limit isn't being hit
LM_COAWAIT window_scroll();
// Get token as string
const auto str = state->vocab.id_to_token[id];
const std::string_view str = state->vocab.id_to_token[id];
// Append string to function result
fres.append(str);
@ -210,12 +213,11 @@ public:
// TODO: Respect batch size
std::vector<int> batch(state->tokens.begin()+state->tokens.size()-1, state->tokens.begin()+state->tokens.size());
if (!mpt_eval(state->model, params.n_threads, state->tokens.size()-1, batch, state->logits, state->mem_per_token)) {
printf("Stopping due to eval error (%d)\n", id);
LM_COTHROW("Failed to evaluate new tokens", "");
}
// Tick
if (on_tick && !on_tick(str.c_str())) abort = true;
if (on_tick && !on_tick(str.data())) abort = true;
else if (!LM_TASKYIELD) abort = true;
}