Update Jarvis implement

2024-01-15 22:52:13 -08:00
parent 5885301c19
commit 0bc2aaf45b
9 changed files with 802 additions and 170 deletions
@@ -160,8 +160,8 @@ class BaseLLMProcess(ABC):
        # Action define in prompt, will be execute after llm compute
        prompt = await self.prepare_prompt(input)
        max_result_token = self.max_token - ComputeKernel.llm_num_tokens(prompt,self.model_name)
-        if max_result_token < MIN_PREDICT_TOKEN_LEN:
-            return LLMResult.from_error_str(f"prompt too long,can not predict")
+        #if max_result_token < MIN_PREDICT_TOKEN_LEN:
+        #    return LLMResult.from_error_str(f"prompt too long,can not predict")
        
        task_result: ComputeTaskResult = await (ComputeKernel.get_instance().do_llm_completion(
                prompt,