Fix text input for TRT-LLM (#151)

triton-inference-server · Oct 24, 2024 · fc0a8bf · fc0a8bf
1 parent 723b5f9
commit fc0a8bf
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 5 deletions.
diff --git a/genai-perf/genai_perf/inputs/converters/tensorrtllm_converter.py b/genai-perf/genai_perf/inputs/converters/tensorrtllm_converter.py
@@ -58,7 +58,7 @@ def convert(
 
                 payload = {
                     "model": model_name,
-                    "text_input": text,
+                    "text_input": [text],
                     "max_tokens": [DEFAULT_TENSORRTLLM_MAX_TOKENS],  # default
                 }
                 self._add_request_params(payload, config)

diff --git a/genai-perf/tests/test_triton_tensorrtllm_converter.py b/genai-perf/tests/test_triton_tensorrtllm_converter.py
@@ -72,12 +72,12 @@ def test_convert_default(self):
             "data": [
                 {
                     "model": "test_model",
-                    "text_input": "text input one",
+                    "text_input": ["text input one"],
                     "max_tokens": [DEFAULT_TENSORRTLLM_MAX_TOKENS],
                 },
                 {
                     "model": "test_model",
-                    "text_input": "text input two",
+                    "text_input": ["text input two"],
                     "max_tokens": [DEFAULT_TENSORRTLLM_MAX_TOKENS],
                 },
             ]
@@ -109,15 +109,15 @@ def test_convert_with_request_parameters(self):
             "data": [
                 {
                     "model": "test_model",
-                    "text_input": "text input one",
+                    "text_input": ["text input one"],
                     "ignore_eos": [True],
                     "max_tokens": [1234],
                     "stream": [True],
                     "additional_key": ["additional_value"],
                 },
                 {
                     "model": "test_model",
-                    "text_input": "text input two",
+                    "text_input": ["text input two"],
                     "ignore_eos": [True],
                     "max_tokens": [1234],
                     "stream": [True],