From a13ba22a3c24a9833685445da0ad0d5678bcdacc Mon Sep 17 00:00:00 2001 From: crasm Date: Mon, 11 Dec 2023 17:53:02 -0500 Subject: [PATCH] llama : document logits_all deprecation llama_context_params.logits_all is a parameter for controlling llama_eval. This documents that logits_all should not be used with llama_decode and llama_batch. --- llama.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/llama.h b/llama.h index b1f5fca624d69..45a65cacb7bb8 100644 --- a/llama.h +++ b/llama.h @@ -216,7 +216,7 @@ extern "C" { // Keep the booleans together to avoid misalignment during copy-by-value. bool mul_mat_q; // if true, use experimental mul_mat_q kernels (DEPRECATED - always true) - bool logits_all; // the llama_eval() call computes all logits, not just the last one + bool logits_all; // the llama_eval() call computes all logits, not just the last one (DEPRECATED - set llama_batch.logits instead) bool embedding; // embedding mode only bool offload_kqv; // whether to offload the KQV ops (including the KV cache) to GPU };