Skip to content

Add a prefill() method for text llm runner #14493

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

Draft
wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

Add a prefill() method for text llm runner #14493

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
beb1784
Add a prefill() method for text llm runner
kirklandsign Sep 22, 2025
3d27fa9
Merge remote-tracking branch 'origin/main' into android-use-prefill-api
kirklandsign Sep 23, 2025
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
22 changes: 22 additions & 0 deletions extension/llm/runner/text_llm_runner.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -217,6 +217,28 @@ Error TextLLMRunner::generate(
return Error::Ok;
}

Error TextLLMRunner::prefill(
const std::string& prompt,
const GenerationConfig& config) {
if (!is_loaded()) {
ET_CHECK_OK_OR_RETURN_ERROR(load());
}

::tokenizers::Result<std::vector<uint64_t>> encode_res = tokenizer_->encode(
prompt,
/*bos=*/config.num_bos,
/*eos=*/config.num_eos);

ET_CHECK_TK_OK_OR_RETURN_ERROR(
encode_res.error(), "Failed to encode prompt %s", prompt.c_str());

// encode the (string) prompt into tokens sequence
std::vector<uint64_t> prompt_tokens = encode_res.get();
auto prefill_res = text_prefiller_->prefill(prompt_tokens, pos_);
ET_CHECK_OK_OR_RETURN_ERROR(prefill_res.error());
return Error::Ok;
}

Error TextLLMRunner::warmup(const std::string& prompt, int32_t max_new_tokens) {
// Create a GenerationConfig for warmup
GenerationConfig config{
Expand Down
11 changes: 11 additions & 0 deletions extension/llm/runner/text_llm_runner.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,6 +101,17 @@ class ET_EXPERIMENTAL TextLLMRunner : public IRunner {
std::function<void(const std::string&)> token_callback = {},
std::function<void(const Stats&)> stats_callback = {}) override;

/**
* Prefill text inputs, for example to reload chat history.
* @param prompt Text prompt to prefill.
* @param config Configuration parameters for text generation (e.g.,
* max_new_tokens, temperature)
* @return The error code. KV cache position is tracked internally in pos_.
*/
::executorch::runtime::Error prefill(
const std::string& prompt,
const GenerationConfig& config);

/**
* @brief Warms up the model with a sample prompt
*
Expand Down
Loading