I get the impression that LAM is running multiple times in parallel, especially when asking follow up questions. At least from observing at my r1 with generative UI, it certainly feels like it. It continuously outputs similar things and takes several loops to get to a next step.
2 Likes
I confirm the behavior. I dod some testing asking the R1 to wirte some code and after 30 seconds, it looks like it starting over, overwirtting what ever is already being written.
3 Likes
Good to know that I am not alone!
I wonder whether this has something to do with usage of generative UI (which is generated with a bit of delay)?
@simon, @rabbit, can somebody look into this, try to reproduce, and ideally fix?
I am open to do a video call and whatever needed to demonstrate the behavior and work with you on getting it fixed!
1 Like
Currently, the problems have improved, and Iām not able to reproduce or currently not experiencing the issue anymore. It works much better now. So maybe it was just a hiccup, and I will reply again if the issue occurs again for me.
2 Likes
Still there, i got the issue again by having it write an html code in the website htmleditor.gitlab.io
After 30 seconds, it will ovewrite itself with a second iteration.
1 Like