huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-06-19 15:52:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

correctly forward back the log probabilities

Browse Source
This commit is contained in:
Morgan Funtowicz 2024-07-17 22:33:10 +00:00
parent bcb96feea6
commit 0212b1774a
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
backends/trtllm/src/ffi.cpp
View File

@ -44,12 +44,12 @@ size_t huggingface::tgi::backends::TensorRtLlmBackendImpl::StreamTokens(
const auto token = decoded.outputTokenIds[0][0]; const auto token = decoded.outputTokenIds[0][0];
const auto isFinal = decoded.isFinal; const auto isFinal = decoded.isFinal;
// const auto logProb = decoded.logProbs.value()[0][0]; const auto logProb = decoded.logProbs.value()[0][0];
++numTokens; ++numTokens;
SPDLOG_DEBUG(FMT_STRING("\tStreamTokens -> {:d} {:.2f} (final = {})"), token, logProb, isFinal); SPDLOG_DEBUG(FMT_STRING("\tStreamTokens -> {:d} {:.2f} (final = {})"), token, logProb, isFinal);
callback(std::move(ctx), token, 1.0, isFinal); callback(std::move(ctx), token, logProb, isFinal);
SPDLOG_DEBUG("\tStreamTokens -> Post callback"); SPDLOG_DEBUG("\tStreamTokens -> Post callback");
} else { } else {
// TODO : Return rest::Result with error // TODO : Return rest::Result with error