Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
update info to run cuda 11.8 version
  • Loading branch information
gorkemgoknar authored Oct 30, 2023
1 parent 2413585 commit 634e257
Showing 1 changed file with 8 additions and 7 deletions.
15 changes: 8 additions & 7 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -8,18 +8,14 @@ To run a pre-built container (CUDA 11.7):
$ docker run --gpus=all -e COQUI_TOS_AGREED=1 --rm -p 8000:80 ghcr.io/coqui-ai/xtts-streaming-server:latest
```

CUDA 11.8 version (for newer cards, tested on 4060)
CUDA 11.8 version (for newer cards, tested on 4060 and L4 instance)
```bash
$ docker run --gpus=all -e COQUI_TOS_AGREED=1 --rm -p 8000:80 ghcr.io/coqui-ai/xtts-streaming-server:latest-cuda118
```

If you have already downloaded v1.1 model and like to use this server, and using Ubuntu, change your /home/YOUR_USER_NAME
```bash
$ docker run \
-v /home/YOUR_USER_NAME/.local/share/tts/tts_models--multilingual--multi-dataset--xtts_v1.1:/root/.local/share/tts/tts_models--multilingual--multi-dataset--xtts_v1.1 \
--env NVIDIA_DISABLE_REQUIRE=1 # When set 1 cuda 11.7 driver will run this container
--gpus=all -e COQUI_TOS_AGREED=1
--rm -p 8000:80 ghcr.io/coqui-ai/xtts-streaming-server:latest-cuda118`
$ docker run -v /home/YOUR_USER_NAME/.local/share/tts/tts_models--multilingual--multi-dataset--xtts_v1.1:/root/.local/share/tts/tts_models--multilingual--multi-dataset--xtts_v1.1 --env NVIDIA_DISABLE_REQUIRE=1 --gpus=all -e COQUI_TOS_AGREED=1 --rm -p 8000:80 ghcr.io/coqui-ai/xtts-streaming-server:latest-cuda118`
```
Setting the `COQUI_TOS_AGREED` environment variable to `1` indicates you have read and agreed to
the terms of the [CPML license](https://coqui.ai/cpml).
Expand All @@ -36,12 +32,17 @@ $ python test_streaming.py

## Building the container

1. To build the Docker container:
1. To build the Docker container (Pytorch 2.01 Cuda 11.7) :

```bash
$ cd server
$ docker build -t xtts-stream .
```
For Pytorch 2.1 and CUDA 11.8 version (when running set NVIDIA_DISABLE_REQUIRE=1 if you have Cuda < 11.8 drivers)
```bash
$ cd server
# docker build -t xtts-stream . -f Dockerfile.cuda118
```
2. Run the server container:

```bash
Expand Down

0 comments on commit 634e257

Please sign in to comment.