I’m having trouble getting the bot to show videos unless specifically asking for a video. I have nothing in custom prompt.
Also, in instructions, does it matter which language i use to describe the video? I’m thinking i’m providing instructions in the main language of the use case to increase key-word recognition.
According to OpenAI, English is the best language. However, if users are typing keywords in another language, I would try to include those within the video description.
Could you send me what you have in the video dataset? That way, I can take a look and advise you.
I have a similar experience - I really have to force the bot to show me video. Even when I include the keywords in the chat. It seems reluctant to offer up the video first and second time around in a conversation. This is despite having the keywords several times and explanations entered in the video description.
In a test conversation I just carrieed out - it said it didn’t have any video multiple times - referred me to my website then actually displayed the video below the text ???
Hi Frank. I have just checked your dashboard, and I saw that you have around 164 datasets. I guess sometimes the videos are not showing because you have a lot of content, and I’m sure there is overlapping or duplication. So probably, sometimes the same keywords are triggering another video, or it matches with a higher score another document. The way you have to describe the videos should be focused on what the user might say to trigger those specific videos. It’s good that you enter a description, but all that description will likely create confusion if you have too many videos.
hi Simone, thanks for your reply much. Any suggestions or can you point me towards any KB article to maximise the efficiency of my bot in its assessment and yes of my data set - any specific advice on a prompt modification et cetera that would help the display a video, if a relevant video is present in the dataset