# AI use cases for Textgen (oobabooga/textgen) Repo: [oobabooga/textgen](https://github.com/oobabooga/textgen) | Use cases: 3 | Page: https://dedupedai.com/use-cases/oobabooga-textgen/ TextGen (formerly text-generation-webui) is an open-source desktop app and web UI for running local LLMs with no telemetry. It supports llama.cpp GGUF models plus Transformers and ExLlamaV3, chat and instruct modes, vision, file attachments, tool-calling with MCP support, an OpenAI and Anthropic compatible API, LoRA training and an image generation tab, all 100% offline. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run a private ChatGPT-style app on your machine: https://dedupedai.com/use-cases/oobabooga-textgen/run-a-private-local-chatgpt-with-textgen/ - Use your local model as an OpenAI-compatible API: https://dedupedai.com/use-cases/oobabooga-textgen/use-textgen-as-a-local-openai-compatible-api/ - Fine-tune a LoRA on your own chats or text: https://dedupedai.com/use-cases/oobabooga-textgen/fine-tune-a-lora-with-textgen-training-tab/ ## How to run a private ChatGPT-style app on your machine with Textgen Repo: [oobabooga/textgen](https://github.com/oobabooga/textgen) | Difficulty: Beginner | Time: 20 minutes | Basis: From the README Install the portable web UI and chat with a GGUF model fully offline. **Who it is for:** Anyone who wants AI chat without sending data to a cloud service. **The problem:** Cloud chat apps see everything you type. **The outcome:** A local chat UI in your browser with your own model. ### Steps 1. Run `git clone https://github.com/oobabooga/textgen` and cd into Textgen 2. Create a venv with `python -m venv venv` and activate it 3. Run `pip install -r requirements/portable/requirements.txt --upgrade` 4. Put a GGUF file in user_data/models and launch with `python server.py --portable --api --auto-launch` ### Prompt to paste into your agent ```text Install oobabooga/textgen with the manual portable venv steps from its README, download a small GGUF model into user_data/models, and launch it with `python server.py --portable --api --auto-launch`. ``` ### Caveat The easiest path is the portable desktop build from the releases page: download, unzip, double-click Textgen. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/oobabooga-textgen/run-a-private-local-chatgpt-with-textgen/ ## How to use your local model as an OpenAI-compatible API with Textgen Repo: [oobabooga/textgen](https://github.com/oobabooga/textgen) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Start TextGen with --api and point existing OpenAI or Anthropic client code at it. **Who it is for:** A developer who wants to test apps against a local model for free. **The problem:** API bills and rate limits slow down local development. **The outcome:** A local endpoint that speaks the OpenAI and Anthropic API formats, with tool calling. ### Steps 1. Install TextGen with the portable venv steps 2. Launch with `python server.py --portable --api --auto-launch` 3. Point your OpenAI-compatible client at the local server 4. Try tool-calling with a custom .py tool or an MCP server ### Prompt to paste into your agent ```text Start oobabooga/textgen with `python server.py --portable --api --auto-launch`, then change to use the local OpenAI-compatible API and run its tests. ``` ### Caveat This card is written from the oobabooga/textgen README, not from a test run by Deduped. Before you start, make sure you have Windows, Linux or macOS. Check the repo for newer instructions before you rely on these steps. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/oobabooga-textgen/use-textgen-as-a-local-openai-compatible-api/ ## How to fine-tune a LoRA on your own chats or text with Textgen Repo: [oobabooga/textgen](https://github.com/oobabooga/textgen) | Difficulty: Advanced | Time: 2 hours | Basis: From the README Use the full install and the Training tab to train a LoRA, with resume support. **Who it is for:** A hobbyist who wants a model that writes in their style. **The problem:** Fine-tuning usually means writing training code. **The outcome:** A LoRA adapter trained on your dataset. ### Steps 1. Clone the repo and run the startup script for your OS, such as start_linux.sh, and select your GPU vendor 2. Open http://127.0.0.1:7860 in your browser 3. Load a Transformers model and open the Training tab 4. Train a LoRA on multi-turn chat or raw text, resuming if interrupted ### Prompt to paste into your agent ```text Help me set up oobabooga/textgen with the one-click installer (start_linux.sh), load a Transformers model, and walk me through training a LoRA on in the Training tab. ``` ### Caveat This card is written from the oobabooga/textgen README, not from a test run by Deduped. Before you start, make sure you have Windows, Linux or macOS. Check the repo for newer instructions before you rely on these steps. Risk, system: The full install needs about 10GB of disk space and downloads PyTorch. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/oobabooga-textgen/fine-tune-a-lora-with-textgen-training-tab/