From 113dd60005f57dbd418e1e0854444b0834841b13 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Fri, 20 Oct 2023 20:42:45 +0300
Subject: [PATCH] server : bach has to be allocated for n_parallel sequences

---
 examples/server/server.cpp | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index 3fa60e0b9..b3add3c40 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -631,7 +631,9 @@ struct llama_server_context
             LOG_TEE(" -> Slot %i - max context: %i\n", slot.id, max_ctx_per_slot);
             slots.push_back(slot);
         }
-        batch = llama_batch_init(n_ctx, 0, 1);
+
+        batch = llama_batch_init(n_ctx, 0, params.n_parallel);
+
         // empty system prompt
         system_prompt = "";
         num_tokens_system = 0;