matt/ollama37
1
0
Fork 0
mirror of https://github.com/dogkeeper886/ollama37.git synced 2025-12-12 00:37:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

Bump llama.cpp to b1999

Browse Source 
This requires an upstream change to support graceful termination,
carried as a patch.
This commit is contained in:
Daniel Hiltgen
2024-01-29 12:58:17 -08:00
parent 78a48de804
commit 72b12c3be7
4 changed files with 130 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
llm/patches/01-cache.diff
View File

@@ -1,8 +1,8 @@
diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index 0462fbd2..4fa7b57f 100644
index a48582ad..9fffffd8 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -1857,12 +1857,6 @@ struct llama_server_context
@@ -1564,12 +1564,6 @@ struct llama_server_context
LOG_TEE("slot %d : in cache: %i tokens | to process: %i tokens\n", slot.id, slot.n_past, slot.num_prompt_tokens_processed);
}
@@ -15,8 +15,8 @@ index 0462fbd2..4fa7b57f 100644
if (slot.n_past == slot.num_prompt_tokens && slot.n_past > 0)
{
// we have to evaluate at least 1 token to generate logits.
@@ -1870,6 +1864,12 @@ struct llama_server_context
slot.n_past--;
@@ -1581,6 +1575,12 @@ struct llama_server_context
}
}
+ LOG_TEE("slot %d : kv cache rm - [%d, end)\n", slot.id, (int) system_tokens.size() + slot.n_past);