Searched refs:mLlamaModelChunks (Results 1 – 2 of 2) sorted by relevance
63 mLlamaModelChunks.push_back(std::move(llamaChunk)); in Initialize()67 auto& modelChunk = mLlamaModelChunks[i]; in Initialize()69 const auto& prevModelChunk = mLlamaModelChunks[i - 1]; in Initialize()84 const auto& tokenEmbInput = mLlamaModelChunks.front()->GetInputBuffer(); in Initialize()89 for (auto& llamaChunk : mLlamaModelChunks) { in Release()92 mLlamaModelChunks.clear(); in Release()99 const auto status = mLlamaModelChunks[chunkIdx]->HotSwapModel(batchSize); in SwapModel()106 for (size_t i = 0; i < mLlamaModelChunks.size(); i++) in SwapModel()108 for (size_t i = 0; i < mLlamaModelChunks.size(); i++) in SwapModel()115 for (auto& modelChunk : mLlamaModelChunks) { in Reset()[all …]
52 std::vector<std::unique_ptr<ModelChunk>> mLlamaModelChunks; variable