From fa6fa730b5b1f47e8ac9ebbd827313267731c875 Mon Sep 17 00:00:00 2001 From: -LAN- Date: Mon, 24 Mar 2025 19:11:33 +0800 Subject: [PATCH] feat: implement NotImplementedError for token counting in LLMs and reintroduce disabled token count method Signed-off-by: -LAN- --- .../model_providers/__base/large_language_model.py | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/api/core/model_runtime/model_providers/__base/large_language_model.py b/api/core/model_runtime/model_providers/__base/large_language_model.py index 9c7465536e..715af3c28d 100644 --- a/api/core/model_runtime/model_providers/__base/large_language_model.py +++ b/api/core/model_runtime/model_providers/__base/large_language_model.py @@ -553,8 +553,7 @@ if you are not sure about the structure. :param tools: tools for tool calling :return: """ - # Disable the token count in LLMs for profermance testing. - return 0 + raise NotImplementedError def enforce_stop_tokens(self, text: str, stop: list[str]) -> str: """Cut off the text as soon as any stop words occur.""" @@ -915,3 +914,7 @@ if you are not sure about the structure. filtered_model_parameters[parameter_name] = parameter_value return filtered_model_parameters + + def _get_num_tokens_by_gpt2(self, text: str) -> int: + # Disable the token count in LLMs for profermance testing. + return 0