text-generation-webui/README.md

164 lines
8.5 KiB
Markdown
Raw Normal View History

2023-01-16 14:31:13 +01:00
# Text generation web UI
2022-12-21 20:49:30 +01:00
2023-01-16 14:07:00 +01:00
A gradio web UI for running large language models like gpt-j-6B, gpt-neo, opt, galactica, and pygmalion on your own computer.
2022-12-21 20:49:30 +01:00
Its goal is to become the [AUTOMATIC1111/stable-diffusion-webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) of text generation.
2022-12-21 17:17:06 +01:00
2023-01-26 20:17:15 +01:00
[[Try it on Google Colab]](https://colab.research.google.com/github/oobabooga/AI-Notebooks/blob/main/Colab-TextGen-GPU.ipynb) -> Updated January 24th, responses are 100x better now <-
2023-01-22 23:29:24 +01:00
2023-01-22 21:28:51 +01:00
|![Image1](https://github.com/oobabooga/screenshots/raw/main/qa.png) | ![Image2](https://github.com/oobabooga/screenshots/raw/main/cai3.png) |
2023-01-15 03:25:05 +01:00
|:---:|:---:|
2023-01-15 08:27:22 +01:00
|![Image3](https://github.com/oobabooga/screenshots/raw/main/gpt4chan.png) | ![Image4](https://github.com/oobabooga/screenshots/raw/main/galactica.png) |
2022-12-21 18:04:51 +01:00
2023-01-07 22:40:51 +01:00
## Features
* Switch between different models using a dropdown menu.
* Notebook mode that resembles OpenAI's playground.
2023-01-22 05:30:55 +01:00
* Chat mode for conversation and role playing.
2023-01-23 14:02:35 +01:00
* Generate nice HTML output for GPT-4chan.
* Generate Markdown output for [GALACTICA](https://github.com/paperswithcode/galai), including LaTeX support.
2023-01-29 07:13:22 +01:00
* Support for [Pygmalion](https://huggingface.co/models?search=pygmalionai/pygmalion) and custom characters in JSON or TavernAI Character Card formats ([see the FAQ](https://github.com/oobabooga/text-generation-webui/wiki/Pygmalion-chat-model-FAQ)).
2023-01-23 14:02:35 +01:00
* Stream the text output in real time.
2023-01-07 22:40:51 +01:00
* Load parameter presets from text files.
2023-01-11 03:30:41 +01:00
* Load large models in 8-bit mode.
2023-01-23 14:02:35 +01:00
* Split large models across your GPU(s), CPU, and disk.
2023-01-09 15:11:05 +01:00
* CPU mode.
2023-01-11 20:51:32 +01:00
* Get responses via API.
2023-01-22 23:29:24 +01:00
* Works on Google Colab ([see the full guide](https://github.com/oobabooga/text-generation-webui/wiki/Running-on-Colab)).
2023-01-07 22:40:51 +01:00
2022-12-21 17:17:06 +01:00
## Installation
Open a terminal and copy and paste these commands one at a time ([install conda](https://docs.conda.io/en/latest/miniconda.html) first if you don't have it already):
2023-01-13 05:29:36 +01:00
```
conda create -n textgen
conda activate textgen
conda install torchvision torchaudio pytorch-cuda=11.7 git -c pytorch -c nvidia
2023-01-20 04:20:35 +01:00
git clone https://github.com/oobabooga/text-generation-webui
cd text-generation-webui
pip install -r requirements.txt
2023-01-20 04:20:35 +01:00
```
The third line assumes that you have an NVIDIA GPU.
* If you have an AMD GPU, you should install the ROCm version of pytorch instead.
2023-01-23 17:41:50 +01:00
* If you are running in CPU mode, you just need the standard pytorch and should replace the third command with this one:
2023-01-13 05:29:36 +01:00
```
conda install pytorch torchvision torchaudio git -c pytorch
2023-01-13 05:29:36 +01:00
```
2022-12-21 17:17:06 +01:00
2023-01-20 01:08:26 +01:00
Once you have completed these steps, you should be able to start the web UI. However, you will first need to download a model.
2022-12-21 18:37:50 +01:00
## Downloading models
2023-01-16 15:19:23 +01:00
Models should be placed under `models/model-name`. For instance, `models/gpt-j-6B` for [GPT-J 6B](https://huggingface.co/EleutherAI/gpt-j-6B/tree/main).
2022-12-21 18:37:50 +01:00
2023-01-15 07:35:57 +01:00
#### Hugging Face
2022-12-21 18:37:50 +01:00
2023-01-15 07:35:57 +01:00
[Hugging Face](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads) is the main place to download models. These are some noteworthy examples:
2022-12-21 18:37:50 +01:00
2023-01-16 15:19:23 +01:00
* [GPT-J 6B](https://huggingface.co/EleutherAI/gpt-j-6B/tree/main)
* [GPT-Neo](https://huggingface.co/models?pipeline_tag=text-generation&sort=downloads&search=eleutherai+%2F+gpt-neo)
2023-01-07 23:19:55 +01:00
* [OPT](https://huggingface.co/models?search=facebook/opt)
* [GALACTICA](https://huggingface.co/models?search=facebook/galactica)
2023-01-06 04:13:26 +01:00
* [\*-Erebus](https://huggingface.co/models?search=erebus)
2023-01-13 13:33:38 +01:00
* [Pygmalion](https://huggingface.co/models?search=pygmalion)
2023-01-06 04:13:26 +01:00
2023-01-23 14:02:35 +01:00
You can automatically download a model from HF using the script `download-model.py`. Its usage is very simple:
2023-01-07 00:00:58 +01:00
python download-model.py organization/model
For instance:
python download-model.py facebook/opt-1.3b
2023-01-23 14:02:35 +01:00
If you want to download a model manually, note that all you need are the json, txt, and pytorch\*.bin files. The remaining files are not necessary.
2023-01-09 20:19:57 +01:00
#### GPT-4chan
2022-12-21 18:37:50 +01:00
2023-01-09 20:19:57 +01:00
[GPT-4chan](https://huggingface.co/ykilcher/gpt-4chan) has been shut down from Hugging Face, so you need to download it elsewhere. You have two options:
2022-12-21 18:37:50 +01:00
* Torrent: [16-bit](https://archive.org/details/gpt4chan_model_float16) / [32-bit](https://archive.org/details/gpt4chan_model)
* Direct download: [16-bit](https://theswissbay.ch/pdf/_notpdf_/gpt4chan_model_float16/) / [32-bit](https://theswissbay.ch/pdf/_notpdf_/gpt4chan_model/)
2023-01-13 05:39:09 +01:00
The 32-bit version is only relevant if you intend to run the model in CPU mode. Otherwise, you should use the 16-bit version.
2023-01-13 05:37:12 +01:00
After downloading the model, follow these steps:
1. Place the files under `models/gpt4chan_model_float16` or `models/gpt4chan_model`.
2023-01-16 15:19:23 +01:00
2. Place GPT-J 6B's config.json file in that same folder: [config.json](https://huggingface.co/EleutherAI/gpt-j-6B/raw/main/config.json).
3. Download GPT-J 6B under `models/gpt-j-6B`:
2023-01-12 03:01:50 +01:00
2023-01-12 03:04:06 +01:00
```
python download-model.py EleutherAI/gpt-j-6B
```
2023-01-08 03:13:43 +01:00
2023-01-16 15:19:23 +01:00
You don't really need all of GPT-J 6B's files, just the tokenizer files, but you might as well download the whole thing. Those files will be automatically detected when you attempt to load GPT-4chan.
2023-01-09 22:58:24 +01:00
#### Converting to pytorch (optional)
2022-12-21 20:49:30 +01:00
2023-01-23 14:05:25 +01:00
The script `convert-to-torch.py` allows you to convert models to .pt format, which is sometimes 10x faster to load to the GPU:
2022-12-21 20:49:30 +01:00
2023-01-08 18:37:43 +01:00
python convert-to-torch.py models/model-name
2022-12-21 20:49:30 +01:00
2023-01-16 14:07:00 +01:00
The output model will be saved to `torch-dumps/model-name.pt`. When you load a new model, the web UI first looks for this .pt file; if it is not found, it loads the model as usual from `models/model-name`.
2022-12-21 20:49:30 +01:00
2023-01-16 14:07:00 +01:00
## Starting the web UI
2022-12-21 17:17:06 +01:00
conda activate textgen
python server.py
2023-01-07 00:00:58 +01:00
Then browse to
`http://localhost:7860/?__theme=dark`
2023-01-07 00:07:58 +01:00
Optionally, you can use the following command-line flags:
2023-01-07 00:00:58 +01:00
2023-01-15 19:40:03 +01:00
| Flag | Description |
|-------------|-------------|
2023-01-15 19:41:36 +01:00
| `-h`, `--help` | show this help message and exit |
| `--model MODEL` | Name of the model to load by default. |
2023-01-16 14:07:00 +01:00
| `--notebook` | Launch the web UI in notebook mode, where the output is written to the same text box as the input. |
| `--chat` | Launch the web UI in chat mode.|
2023-01-25 23:43:30 +01:00
| `--cai-chat` | Launch the web UI in chat mode with a style similar to Character.AI's. If the file `img_bot.png` or `img_bot.jpg` exists in the same folder as server.py, this image will be used as the bot's profile picture. Similarly, `img_me.png` or `img_me.jpg` will be used as your profile picture. |
2023-01-15 19:41:36 +01:00
| `--cpu` | Use the CPU to generate text.|
| `--load-in-8bit` | Load the model with 8-bit precision.|
| `--auto-devices` | Automatically split the model across the available GPU(s) and CPU.|
| `--disk` | If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk. |
| `--disk-cache-dir DISK_CACHE_DIR` | Directory to save the disk cache to. Defaults to `cache/`. |
2023-01-21 04:33:41 +01:00
| `--gpu-memory GPU_MEMORY` | Maximum GPU memory in GiB to allocate. This is useful if you get out of memory errors while trying to generate text. Must be an integer number. |
2023-01-21 07:05:55 +01:00
| `--cpu-memory CPU_MEMORY` | Maximum CPU memory in GiB to allocate for offloaded weights. Must be an integer number. Defaults to 99.|
| `--no-stream` | Don't stream the text output in real time. This improves the text generation performance.|
2023-01-16 20:35:45 +01:00
| `--settings SETTINGS_FILE` | Load the default interface settings from this json file. See `settings-template.json` for an example.|
| `--extensions EXTENSIONS` | The list of extensions to load. If you want to load more than one extension, write the names separated by commas and between quotation marks, "like,this". |
2023-01-21 03:45:16 +01:00
| `--listen` | Make the web UI reachable from your local network.|
2023-01-29 06:54:36 +01:00
| `--listen-port LISTEN_PORT` | The listening port that the server will use. |
2023-01-19 21:31:29 +01:00
| `--share` | Create a public URL. This is useful for running the web UI on Google Colab or similar. |
2023-01-26 06:12:53 +01:00
| `--verbose` | Print the prompts to the terminal. |
2023-01-09 14:58:46 +01:00
2023-01-19 15:25:17 +01:00
Out of memory errors? [Check this guide](https://github.com/oobabooga/text-generation-webui/wiki/Low-VRAM-guide).
2023-01-06 05:33:21 +01:00
## Presets
Inference settings presets can be created under `presets/` as text files. These files are detected automatically at startup.
2023-01-24 00:44:27 +01:00
By default, 10 presets by NovelAI and KoboldAI are included. These were selected out of a sample of 43 presets after applying a K-Means clustering algorithm and selecting the elements closest to the average of each cluster.
2023-01-09 22:12:41 +01:00
## System requirements
2023-01-09 22:18:35 +01:00
Check the [wiki](https://github.com/oobabooga/text-generation-webui/wiki/System-requirements) for some examples of VRAM and RAM usage in both GPU and CPU mode.
2023-01-09 22:12:41 +01:00
2022-12-21 20:52:23 +01:00
## Contributing
2023-01-16 14:07:00 +01:00
Pull requests, suggestions, and issue reports are welcome.
2023-01-09 20:15:54 +01:00
2023-01-15 07:00:59 +01:00
## Credits
2023-01-09 20:15:54 +01:00
2023-01-15 08:29:19 +01:00
- NovelAI and KoboldAI presets: https://github.com/KoboldAI/KoboldAI-Client/wiki/Settings-Presets
2023-01-25 14:17:55 +01:00
- Pygmalion preset, code for early stopping in chat mode: https://github.com/PygmalionAI/gradio-ui/
2023-01-16 15:23:45 +01:00
- Verbose preset: Anonymous 4chan user.
- Gradio dropdown menu refresh button: https://github.com/AUTOMATIC1111/stable-diffusion-webui