Docker Container for <a href=https://github.com/ebolam/KoboldAI>https://github.com/ebolam/KoboldAI
Volumes:
/content/settings -> directory that will contain user settings
/content/models -> directory that contains any models that were downloaded
/content/stories -> directory that your stories will be saved to
/content/softprompts -> directory that houses your soft prompt files
/content/userscripts -> directory that houses your userscripts
Environmental Variables:
| Variable | Default | Description |
|---|---|---|
| remote | False | Optimizes KoboldAI for Remote Play |
| noaimenu | False | Disables the ability to select the AI |
| ngrok | False | Optimizes KoboldAI for Remote Play using Ngrok |
| localtunnel | False | Optimizes KoboldAI for Remote Play using Localtunnel |
| host | False | Optimizes KoboldAI for Remote Play without using a proxy service |
| port port | 5000 | Specify the port on which the application will be joinable |
| aria2_port ARIA2_PORT | 6799 | Specify the port on which aria2's RPC interface will be open if aria2 is installed (defaults to 6799) |
| model model | ReadOnly | Specify the Model Type to skip the Menu |
| path path | "" | Specify the Path for local models (For model NeoCustom or GPT2Custom) |
| revision revision | "" | Specify the model revision for huggingface models (can be a git branch/tag name or a git commit hash) |
| cpu | False | By default unattended launches are on the GPU use this option to force CPU usage. |
| breakmodel_gpulayers | "" | If using a model that supports hybrid generation, this is a comma-separated list that specifies how many layers to put on each GPU device. For example to put 8 layers on device 0, 9 layers on device 1 and 11 layers on device 2, use --beakmodel_gpulayers 8,9,11 |
| breakmodel_disklayers | "" | If using a model that supports hybrid generation, this is the number of layers to put in disk cache. |
| override_delete | False | Deleting stories from inside the browser is disabled if you are using --remote and enabled otherwise. Using this option will instead allow deleting stories if using --remote and prevent deleting stories otherwise. |
| override_rename | False | Renaming stories from inside the browser is disabled if you are using --remote and enabled otherwise. Using this option will instead allow renaming stories if using --remote and prevent renaming stories otherwise |
| configname | "" | Force a fixed configuration name to aid with config management. |
| colab | False | Optimize for Google Colab. |
| nobreakmodel | False | Disables Breakmodel support completely. |
| unblock | False | Unblocks the KoboldAI port to be accessible from other machines without optimizing for remote play (It is recommended to use --host instead) |
| quiet | False | If present will suppress any story related text from showing on the console |
| no_aria2 | False | Prevents KoboldAI from using aria2 to download huggingface models more efficiently, in case aria2 is causing you issues |
| lowmem | False | Extra Low Memory loading for the GPU, slower but memory does not peak to twice the usage |
| savemodel | False | Saves the model to the models folder even if --colab is used (Allows you to save models to Google Drive) |
Content type
Image
Digest
sha256:12e6bd025…
Size
10.7 GB
Last updated
about 3 years ago
docker pull ebolam/koboldai