From 314e729347defd9851e857f78084160c5786a7d8 Mon Sep 17 00:00:00 2001 From: Tim Neumann Date: Mon, 25 May 2026 09:29:28 +0200 Subject: [PATCH] llama : document that only one on-device state can be saved per sequence (#23520) --- include/llama.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/include/llama.h b/include/llama.h index 75095b22d0..e8374c53b7 100644 --- a/include/llama.h +++ b/include/llama.h @@ -874,7 +874,8 @@ extern "C" { // work only with partial states, such as SWA KV cache or recurrent cache (e.g. Mamba) #define LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY 1 -// keeps the tensor data on device buffers (i.e. not accessible in host memory, but faster save/load) +// Keeps the tensor data on device buffers (i.e. not accessible in host memory, but faster save/load). +// Getting the state for a seq_id with this flag invalidates all prior states gotten for that seq_id with this flag. #define LLAMA_STATE_SEQ_FLAGS_ON_DEVICE 2 typedef uint32_t llama_state_seq_flags;