Metadata-Version: 2.4
Name: quivr-mempalace-rag-ui
Version: 0.1.0
Summary: A Gradio RAG application combining Quivr document search with MemPalace memory
Author: Gaurav Sarma
License-Expression: MIT
Requires-Python: >=3.11
Description-Content-Type: text/markdown
Requires-Dist: quivr-core<0.1,>=0.0.33
Requires-Dist: quivr-mempalace>=0.1.0
Requires-Dist: gradio<7,>=5
Requires-Dist: langchain-core<0.4,>=0.3
Requires-Dist: langchain-groq<1,>=0.3.5
Requires-Dist: langchain-nvidia-ai-endpoints<1,>=0.3.19
Requires-Dist: langfuse<3,>=2.57
Requires-Dist: packaging<25,>=23.2
Requires-Dist: python-dotenv>=1.0

# Combined Quivr + MemPalace RAG App

This app combines the local Quivr Core fork with the MemPalace integration and
a Gradio browser UI.

The setup script uses `uv` and installs both local fork-style checkouts:
`forks\quivr` and `forks\mempalace`.

## Install as a pip package

After publishing the package, users can install and run it with:

```cmd
python -m pip install quivr-mempalace-rag-ui
quivr-mempalace-rag-ui
```

The package automatically installs its `quivr-mempalace` dependency. Both
packages must be published before the one-command install works for new users.

## Install from CMD

```cmd
cd /d C:\Users\GauravSarma\Downloads\quivr-main\quivr-main\forks\quivr\examples\mempalace_rag_ui
setup.cmd
notepad .env
```

## Start the app

```cmd
cd /d C:\Users\GauravSarma\Downloads\quivr-main\quivr-main\forks\quivr\examples\mempalace_rag_ui
run_app.cmd
```

## Build and publish from CMD

```cmd
cd /d C:\Users\GauravSarma\Downloads\quivr-main\quivr-main\forks\quivr\examples\mempalace_rag_ui
build_all.cmd
publish_all.cmd
```

Set your PyPI token before publishing:

```cmd
set UV_PUBLISH_TOKEN=pypi-your-token
publish_package.cmd
```

Open `http://127.0.0.1:7862`.

1. Choose Groq or NVIDIA NIM.
2. Upload one or more documents.
3. Click **Index documents**.
4. Click **Recall memories** to test MemPalace alone.
5. Click **Ask** to run Quivr document RAG plus MemPalace memory context.

The app uses NVIDIA embeddings for document indexing. Hosted embeddings require
`NVIDIA_API_KEY`; a self-hosted NVIDIA embedding NIM can be selected with
`NVIDIA_EMBEDDING_BASE_URL`. Hugging Face support remains available in the
standalone integration package through its optional `huggingface` extra.
