CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-05 09:36:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : unified KV cache + batch inference API

Browse Source
This commit is contained in:
Georgi Gerganov
2023-09-18 10:08:22 +03:00
parent fad56936d4
commit d29e76937c
10 changed files with 315 additions and 236 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/perplexity/perplexity.cpp
View File

@@ -400,7 +400,7 @@ results_perplexity perplexity(llama_context * ctx, const gpt_params & params) {
return {tokens, ppl, logit_history, prob_history};
}
std::vector<float> hellaswag_evaluate_tokens(llama_context * ctx, const std::vector<int>& tokens, int n_past, int n_batch,
std::vector<float> hellaswag_evaluate_tokens(llama_context * ctx, const std::vector<int> & tokens, int n_past, int n_batch,
int n_vocab, int n_thread) {
std::vector<float> result;
result.reserve(tokens.size() * n_vocab);