How to quickly "reset" when prototyping with Flowise with an in-memory Vector DB
When prototyping with Flowise, I'll create a flow with a document uploader, an in-memory vector DB, and a chat agent to trigger the flow. When I want to "reset" the prototype, and try something else (fast iteration), I don't know how to clear the data out so I can try something different. (See flow in picture.)
I've tried is changing my data that's to be uploaded, deleting the in-memory database and the file uploader, adding them in again, and clicking save. However the chat agent still responds as the same as before.
What's a fast way to clear the chat flow of previously uploaded data?
What’s not intuitive is that “save” operation needs to be done to regenerate code changes each time AND chat cache needs to be cleared.
To reset a chat flow so you can feed it new data (as shown in the original poster's picture) you need to do the following:
- make the changes wanted to the data file accessed by the file uploader (see attached picture) click "Clear Chat History" icon and click Save icon
- delete in-memory vector store and click Save icon
- add in new in-memory vector store and click Save icon
Then in the next chat, it will check it's cached chat history but it will be blank so then: the new data will flow from the file uploader, and be embedded into the in-memory vector store, and the new data in the vector store will be used to pass along the top "winners" from the DB to the LLM and the chat agent will sort out what it wants to generate.
- Removing strange/special characters from outputs llama 3.1 model
- RuntimeError: Library libcublas.so.11 is not found or cannot be loaded
- Unexpected value passed to LangChain Tool argument
- Error during the compilation of the tokenizers package when trying to install transformers 4.27
- How are the weights of the Mistral models reinitialized in Huggingface?
- Using a text embedding model locally with semantic kernel
- Run Anthropic API in parallel
- How to extract information from a photo of a document
- Running LLMs locally causing this error: The request was canceled due to the configured HttpClient.Timeout of 100 seconds elapsing
- How to pass max_token_to_sample parameter when using boto3 to access AWS bedrock model with Knowledgebase
- Could not parse ModelProto from Meta-Llama-3.1-8B-Instruct/tokenizer.model
- Embedding using the LangChain_AWS is giving None value
- Debugging Dynamic Few Shot Langchain Code
- How to install and run Ollama server in AWS Kubernetes cluster (EKS)?
- What is the exact vocab size of the Mistral-Nemo-Instruct-2407 tokenizer model?
- Error using LlmFactory with "TheBloke/OpenHermes-2.5-Mistral-7B-GGUF" Huggingface
- Vector store created using existing graph for multiple nodes/labels
- TypeError: Expected a Runnable, callable or dict.Instead got an unsupported type: <class 'list'>
- Getting Peft Version Error while Autotrain Finetune on Llama 2
- NER versus LLM to extract name, gender, role and company from text
- Where does anythingLLM desktop save its embedding database files
- How to quantize safetensors model and save it to GGUF with less then q8_0 quntization?
- Ollama with RAG for local utilization to chat with pdf
- How to develop a Generalized RAG Pipeline for Text, Images, and Structured Data
- Error when attempting to add data source to Azure OpenAI api
- NameError: name 'LangchainLLM' is not defined
- How to return used context to answer using Langchain in Python
- LangChain Chat History
- Why Langchain HuggingFaceEmbeddings model dimension is not the same as stated on HuggingFace
- Ollama not saving anything in context