Skip to content

LLM inference in C/C++

C++ ggml

tests : adjust server string regex to also match m2 utlra results (#29648)

Georgi Gerganovc85b92c69c95
diff --git a/tools/server/server-context.cpp b/tools/server/server-context.cpp
index 2036319c13..fbfcbe5125 100644
--- a/tools/server/server-context.cpp
+++ b/tools/server/server-context.cpp
@@ -3155,7 +3155,7 @@ private:
// TODO: support memory-less logits computation
if (slot.task->need_logits() && !llama_get_memory(ctx_tgt)) {
- send_error(slot, "the current context does not logits computation. skipping", ERROR_TYPE_SERVER);
+ send_error(slot, "the current context does not support logits computation. skipping", ERROR_TYPE_SERVER);
slot.release();
return;
}
diff --git a/tools/server/tests/unit/test_completion.py b/tools/server/tests/unit/test_completion.py
index 01732eb163..09482b75ce 100644
--- a/tools/server/tests/unit/test_completion.py
+++ b/tools/server/tests/unit/test_completion.py
@@ -148,7 +148,7 @@ def test_completion_stream_with_openai_library_stops():
if choice.finish_reason is None:
assert choice.text is not None
output_text += choice.text
- assert match_regex("Sure, here's one for[\\s\\S]*", output_text), f'Unexpected output: {output_text}'
+ assert match_regex("Sure, here's one for[\\s\\S]*|Sure! Here's one for you[\\s\\S]*", output_text), f'Unexpected output: {output_text}'
@pytest.mark.parametrize("n_slots", [1, 2])