eMediaLibrary Project

That's some other error

this is locally?

Yes, but I think in the Templates sites it is also not working ok

you could grab the response as a stirng and echo it out
in Ollama Manager

yeah, I'm just debugin it

see what the error is, what model are you testing it

HttpResponseProxy{HTTP/1.1 400 Bad Request [Server: nginx/1.20.1, Date: Fri, 21 Feb 2025 17:36:11 GMT, Content-Type: application/json; charset=utf-8, Content-Length: 61, Connection: keep-alive] ResponseEntityProxy{[Content-Type: application/json; charset=utf-8,Content-Length: 61,Chunked: false]}}

what does your input look like

{ "model": "llama3.2", "stream": false, "messages": [ { "role": "system", "content": "You're a powerful chat assistant inside eMedia - a versatile database. You\nact like a bit of a search assistant. When a user asks for info on a topic, you \nshould look for references in the database on that topic rather than answering off the cuff, and then \nsummarize the results.\n\nThe search will return various types of assets, images, PDFs etc.\n\nBe serious but friendly. \n\n\n If your first search doesn't return anything, try searching again a few times with alternative keywords. \n Only do this 3 times then just let the user know we don't have that info.\n\nThe user has access to a variety of top level tables which are records that contain data. This database contains various kinds of data depending on how the system has been configured.\n\nThis user has these tables in particular :\n\n\t\tWith a NAME of \"Activities\" has an id of ID: \"entityactivity\" \t\t\t\t\t\n and \t\tWith a NAME of \"Categories\" has an id of ID: \"searchcategory\" \t\t\t\t\t\n and \t\tWith a NAME of \"Collections\" has an id of ID: \"librarycollection\" \t\t\t\t\t\n and \t\tWith a NAME of \"Content Creator\" has an id of ID: \"contentcreator\" \t\t\t\t\t\n and \t\tWith a NAME of \"Facial Profiles\" has an id of ID: \"faceprofilegroup\" \t\t\t\t\t\n and \t\tWith a NAME of \"Files\" has an id of ID: \"asset\" \t\t\t\t\t\n and \t\tWith a NAME of \"Media Profile\" has an id of ID: \"socialmediaprofile\" \t\t\t\t\t\n and \t\tWith a NAME of \"Objects\" has an id of ID: \"entityobject\" \t\t\t\t\t\n and \t\tWith a NAME of \"Person\" has an id of ID: \"entityperson\" \t\t\t\t\t\n and \t\tWith a NAME of \"Projects\" has an id of ID: \"entityproject\" \t\t\t\t\t\n and \t\tWith a NAME of \"Users\" has an id of ID: \"user\" \t\t\t\t\t\n\nYou can search-tables in response to the users query. You need to identify the ID of the table in question. \n\nEVERYTIME you search tables you MUST respond with a emediasearchformat.\n" }, { "role": "user", "content": "search for images with flowers " } ] , "tools": [ { "type": "function", "function": 404 on /finder/mediadb/gpt/functiondefs/search-tables.json } ], "function_call": "auto" }

Gotta be some bad json or something

"function": 404 on /finder/mediadb/gpt/functiondefs/search-tables.json

Oh I see a 404

That's the problem, I think you guys were renaming stuff

ok ok yeah I think shakil move those or something, let me find them

Chris change the function name

you just need to match the input in the openai version

"functions":[ ##$gpt.loadInputFromTemplate($context, "/$mediaarchive.getMediaDbId()/gpt/functiondefs/checkweather.json"), #foreach($module in $chatprofile.getEntities()) $context.putPageValue("module",$module) $gpt.loadInputFromTemplate($context, "/$mediaarchive.getMediaDbId()/gpt/functiondefs/search-moduletemplate.json"), #end $gpt.loadInputFromTemplate($context, "/$mediaarchive.getMediaDbId()/gpt/functiondefs/emediasearchformat.json") ],
See this - he changed it to emediasearchformat

but didn't update it in the ollaman API format also
no big deal just change it to match

The inputs are similar but you can't use one input for both, they are different enough apis

but you can reuse the function definitions, they are identical