Harrison Chase

@waleedkadous agree that #1 seems the easiest/highest priority. happy to help with that in any way

some feedback we've gotten from users is theres a lot of inheritance already and it can make it tricky to understand whats going on. although this may save code, i...

fixed! thanks!

yeah - was chatting with @asamant21 about it. step 1 i think is be able to serialize a LLM configuration to json once that is done, we can just create...

yeah i think some flag for whether or not to cache each llm call is probably good, will work on tht

okay so possible modes would be: 1. no cache (do not read or write from cache) 2. cache (read and write from cache) 3. recompute (only write to cache) is...

yup exactly. okay not 100% sure when ill have time to get to this, but seems clear what needs to be done

> I'm happy to look at it too. Any ideas what the interface should look like? Should it be available for all chain types? My use case is summarization, so...

glad to hear! just got released, will tweet about it tmrw!

> I can help with it, always want something like this : ) yay! let me know if you want to bounce any ideas around, otherwise ill keep an eye...