langflow/docs/multiple-vectorstores.md
carlosrcoelho 5d2a29a436 add docs
2023-07-18 14:59:27 -03:00

122 lines
3 KiB
Markdown
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

`VectorStoreRouterAgent` construct an agent that routes between multiple vectorstores.
<br>
![Description](img/single_node/mult_vect.png#only-light){width=50%}
![Description](img/single_node/mult_vect2.png#only-dark){width=50%}
<br>
For more information about [VectorStoreRouterAgent](https://python.langchain.com/en/latest/modules/agents/agent_executors/examples/agent_vectorstore.html?highlight=Router){.internal-link target=\_blank}, check out the LangChain documentation.
---
### ⛓LangFlow example
![Description](img/multiple-vectorstores2.png#only-dark){width=100%}
![Description](img/multiple-vectorstores.png#only-light){width=100%}
<br>
[Download Flow](data/Multiple_vectorstores.json){: .md-button download="Multiple_vectorstores"}
<br>
`TextLoader` loads text from a file.
<br>
[Download txt](data/state_of_the_union.txt){: .md-button download="state-of-the-union"}
<br>
By using `WebBaseLoader`, you can load all text from webpages into a document format that we can use downstream. Web path used:
```txt
https://beta.ruff.rs/docs/faq/
```
`CharacterTextSplitter` implements splitting text based on characters.
Text splitters operate as follows:
- Split the text into small, meaningful chunks (usually sentences).
- Combine these small chunks into larger ones until they reach a certain size (measured by a function).
- Once a chunk reaches the desired size, make it its piece of text and create a new chunk with some overlap to maintain context.
**Separator used**:
```txt
.
```
**Chunk size used**:
```txt
2000
```
**Chunk overlap used**:
```txt
200
```
<br>
The `OpenAIEmbeddings`, wrapper around [OpenAI Embeddings](https://platform.openai.com/docs/guides/embeddings/what-are-embeddings){.internal-link target=\_blank} models. Make sure to get the API key from the LLM provider, in this case [OpenAI](https://platform.openai.com/){.internal-link target=\_blank}.
<br>
`Chroma` vector databases can be used as vectorstores to conduct a semantic search or to select examples, thanks to a wrapper around them.
<br>
`VectorStoreInfo` set information about the vectorstore, such as the name and description.
<br>
**First VectorStoreInfo**
<br>
Name:
```txt
state_of_union_address
```
Description:
```txt
the most recent state of the Union address
```
**Second VectorStoreInfo**
<br>
Name:
```txt
ruff
```
Description:
```txt
Information about the Ruff python linting library
```
<br>
The `VectorStoreRouterToolkit` is a toolkit that allows you to create a `VectorStoreRouter` agent. This allows it to route between vector stores.
<br>
For the example, we used `OpenAI` as the LLM, but you can use any LLM that has an API. Make sure to get the API key from the LLM provider. For example, [OpenAI](https://platform.openai.com/){.internal-link target=\_blank} requires you to create an account to get your API key.
<br>
Check out the [OpenAI](https://platform.openai.com/docs/introduction/overview){.internal-link target=\_blank} documentation to learn more about the API and the options that contain in the node.