watsonx.ai

watsonx.ai

A one-stop, integrated, end- to-end AI development studio

 View Only
  • 1.  Regarding prompt generation

    Posted 10/05/23 09:22 AM

    Hello @team,

    I have tried to generate a 20 lines email output using available LLM models in watsonx ai in the below format

    prompt_input = """generate a 20 lines email escalating rahul for his work"""

    and i got the output as below:

    'rahul, i am escalating you for your work on the saskatchewan project. 
    i am sure you will do a great job. i am sure you will be able to get the project done on time. 
    i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. i am sure you will be able to get the project done on time. 
    i am sure you will be able to get the project done on time. i am sure you will be able to get the project done'

    I have used model_id = "google/flan-t5-xxl" and max_newtokens=1024.

    The output was really shocking ,could you please guide me if iam sing the llm in wrong way.

    Thanks

    Nitesh


    #watsonx.ai

    ------------------------------
    Nitesh Velamuri
    ------------------------------


  • 2.  RE: Regarding prompt generation

    Posted 10/05/23 03:35 PM

    With Greedy decoding, you may need to increase the repetition_penalty to 1.4 or 1.5



    ------------------------------
    Alexei Karve Sr. Technical Staff Member (STSM)
    ------------------------------