what is the custom system prompt argument in chat engine

is there away to build a custom response synthesizer? i want to add it to speak like a salesman,

export const llamaPipeline = async(content) =>{ let OpenEmbedllm = new OpenAIEmbedding({ temperature: 0, apiKey: process.env.OPENAI_SECRET_KEY, model:'text-embedding-3-small', apiVersion:"2023-07-01-preview", }) try{ const resFromAnalyze = await client.query(SELECT overlap,chunk_size
FROM public.${process.env.PG_DB_PREFIX}_prompts;) let pipeline = new IngestionPipeline({transformations:[ // new RemoveSpecialCharacters(), new SimpleNodeParser({ chunkOverlap:200, chunkSize:20}), new KeywordExtractor({llm:azureOpenAillm}), OpenEmbedllm ]}) const nodes = await pipeline.run({documents:[new Document({text:content})]}) console.log(nodes) return nodes }catch(err){ console.log(err) } }

hey, is there a way to handle images in pdf files ?

@kapa.ai how to make extractQuestionsFromNode not consider the metadata of the node

is there splitter , that I can use in my llamapipeline that will know to deal with paragraphs and list of items? and not cut them in the middle?

hey, I am trying to force a chatEngine to talk like salesman so I modified its systemContextPrompt and stated to him to not "invent" stuff and only use its sources, but he ignores it. does anyone has a good wording for forcing it? its using gpt-3.5-turbo

too bad I am using ts seems like this method is not supported there

do i have a way to see the prompts that are being sen with the chatEngine? doing getPrompts() only reveal the systemContextPrompt

in our current flow we take the topk nodes that are returned and then we pass to the llm so I would rank them, is there a built in way to pass all the nodes to the llm and prompt it to rank them based on some context?

is it included in the embedding if I use the llm to embed it? , does it just append it to the node text before embedding the chunk+ title