huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-04-21 23:12:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

(ffi) add missing headers imports

Browse Source
This commit is contained in:
Morgan Funtowicz 2024-08-02 22:17:32 +00:00 committed by Morgan Funtowicz
parent 2883c042ed
commit 33c962ef41
2 changed files with 6 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
backends/trtllm/include/ffi.h
View File

@ -5,7 +5,9 @@
#ifndef TGI_TRTLLM_BACKEND_FFI_H
#define TGI_TRTLLM_BACKEND_FFI_H
ad#include <cmath>
#include <cstddef>
#include <memory>
#include "backend.h"
namespace huggingface::tgi::backends {

6
backends/trtllm/src/ffi.cpp
View File

@ -3,11 +3,13 @@
//
#pragma once
#include <cmath>
#include <algorithm>
#include <exception>
#include <filesystem>
#include <functional>
#include <limits>
#include <iterator>
#include <ranges>
#include <vector>
#include <spdlog/spdlog.h>
@ -28,7 +30,7 @@ uint64_t huggingface::tgi::backends::TensorRtLlmBackendImpl::Submit(
// This will copy all the items from the initial slice
std::vector<int32_t> tokens_(std::make_move_iterator(tokens.begin()), std::make_move_iterator(tokens.end()));
return TensorRtLlmBackend::Submit(
std::move(tokens_), topK, topP, temperature, repetition_penalty, frequency_penalty, seed);
std::move(tokens_), maxNewTokens, topK, topP, temperature, repetition_penalty, frequency_penalty, seed);
}
std::unique_ptr<std::vector<huggingface::tgi::backends::GenerationStep>>