- Update PyTorch installation URL in docker-entrypoint.sh to use the nightly builds with CUDA 12.8
- Change from cu126 to cu128 to match the CUDA version in Dockerfile-cuda128
Update the Docker configuration to use environment variables for database connections instead of hardcoded values:
- Add POSTGRES_HOST environment variable to make database hostname configurable
- Generate DATABASE_URL dynamically in docker-compose.yml using environment variables
- Update pg_isready command in entrypoint script to use POSTGRES_HOST variable
- Update env.example to reflect these changes and explain the transition
This change improves configuration flexibility and makes the deployment more portable across different environments.
This commit adds comprehensive support for displaying AI's thinking process in the transcript UI:
1. Update ThinkingDisplay component:
- Improve rendering of thinking sections with expandable/collapsible UI
- Add proper toggle functionality for showing/hiding thinking sections
- Enhance styling and readability of thinking sections
2. Update AudioPlayer component for Svelte 5 compatibility:
- Convert to use $state for reactive variables
- Refactor to use modern event handling
- Improve waveform rendering and progress tracking
3. Update summary display with markdown rendering support:
- Add proper formatting for AI-generated summaries
- Improve UI for viewing thinking sections in summaries
- Connect thinking display toggle to parent component state
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Updated `.dockerignore`:
- Renamed Dockerfile from `Dockerfile-gpu` to `Dockerfile-cuda128` in ignore list.
- Removed the extra newline for cleaner format.
- Updated GitHub Actions workflows:
- Modified `main-cuda-docker.yml`, `nightly-cuda-docker.yml` to change Dockerfile reference from `Dockerfile-gpu` to `Dockerfile-cuda128`.
- Updated image tags from `-gpu` to `-cuda128`.
- Removed build and verification steps for non-CUDA Docker images from `main-docker.yml` and `nightly-docker.yml`.
- Updated main `Dockerfile`:
- Adjusted installation steps for Node.js dependencies.
- Added removal of `.env` file post-build to keep images clean.
- Updated `README.md` for GPU image build command:
- Changed reference from `Dockerfile-gpu` to `Dockerfile-cuda128`.
- Modified `docker-compose.gpu.yml` and `docker-compose.yml`:
- Updated Dockerfile reference in `docker-compose.gpu.yml` from `Dockerfile-gpu` to `Dockerfile-cuda128`.
- Set default paths for `MODELS_DIR`, `WORK_DIR`, and `AUDIO_DIR` using environment variables and adjusted volume configurations.
- Refactored `docker-entrypoint.sh`:
- Changed virtual environment directory from `/app/deps/` to `/scriberr/`.
- Removed the dependency check marker file logic for a more direct installation approach.
- Updated `.gitignore` and `env.example`:
- Removed `*.env` entry from `.gitignore`.
- Added descriptions for paths in `env.example` for clarity.
These changes streamline the Docker setup, improve environment variable handling, and update documentation for clearer instructions.
- Updated `Dockerfile` to use Ubuntu 24.04 as the base image.
- Created `Dockerfile-gpu` for GPU support using `nvidia/cuda:12.8.0-cudnn-runtime-ubuntu24.04`.
- Moved dependency installations to `docker-entrypoint.sh` for both Python and Node.js.
- Ensured virtual environment and Node.js dependencies are checked and installed during deployment.
- Added logic to handle hardware acceleration (CPU/GPU) in `docker-entrypoint.sh`.
- Update Dockerfile to use Ubuntu 24.04 and improve installation with `apt-get upgrade`.
- Modify Dockerfile to create and use a virtual environment for Python dependencies.
- Introduce ARGs for naming in Dockerfile-gpu and update base image to Ubuntu 22.04.
- Comment out PostgreSQL port mapping in docker-compose.yml for improved security (not exposed by default).
- Update docker-entrypoint.sh to activate the Python virtual environment before app startup.
- Simplify Python package installs in +server.ts by removing version-specific constraints.
Dockerfile-gpu: Changed base image to nvidia/cuda:11.8.0-cudnn8-devel-ubuntu22.04 to enable GPU acceleration with CUDA.
docker-compose.gpu.yml: Modified to build from Dockerfile-gpu and added GPU resource reservations for NVIDIA devices.
docker-entrypoint.sh: Updated to set HARDWARE_ACCEL environment variable to 'cuda' when 'gpu' is specified, defaulting to 'cpu' otherwise.
transcription.ts: Adjusted the transcription service to use the HARDWARE_ACCEL environment variable for device selection, allowing dynamic choice between CPU and GPU.
These changes enable optional GPU acceleration for improved performance and update default configurations to use more widely supported models. Users with NVIDIA GPUs can now leverage CUDA for faster processing.
- Updated Dockerfile to install WhisperX and Diarization models and added npm build step.
- Modified docker-compose.yml to utilize environment variables and provide options for local data storage.
- Enhanced database connection in docker-entrypoint.sh to use environment variables for credentials.
- Added support for HuggingFace API key and diarization model in transcription.ts and transcribe.py.
