One is anyToAny models that requires a VAD model, and one is
wrappedModel that requires as well VAD models along others in the
pipeline.
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Makes the web app honour the `X-Forwarded-Prefix` HTTP request header that may be sent by a reverse-proxy in order to inform the app that its public routes contain a path prefix.
For instance this allows to serve the webapp via a reverse-proxy/ingress controller under a path prefix/sub path such as e.g. `/localai/` while still being able to use the regular LocalAI routes/paths without prefix when directly connecting to the LocalAI server.
Changes:
* Add new `StripPathPrefix` middleware to strip the path prefix (provided with the `X-Forwarded-Prefix` HTTP request header) from the request path prior to matching the HTTP route.
* Add a `BaseURL` utility function to build the base URL, honouring the `X-Forwarded-Prefix` HTTP request header.
* Generate the derived base URL into the HTML (`head.html` template) as `<base/>` tag.
* Make all webapp-internal URLs (within HTML+JS) relative in order to make the browser resolve them against the `<base/>` URL specified within each HTML page's header.
* Make font URLs within the CSS files relative to the CSS file.
* Generate redirect location URLs using the new `BaseURL` function.
* Use the new `BaseURL` function to generate absolute URLs within gallery JSON responses.
Closes#3095
TL;DR:
The header-based approach allows to move the path prefix configuration concern completely to the reverse-proxy/ingress as opposed to having to align the path prefix configuration between LocalAI, the reverse-proxy and potentially other internal LocalAI clients.
The gofiber swagger handler already supports path prefixes this way, see e2d9e9916d/swagger.go (L79)
Signed-off-by: Max Goltzsche <max.goltzsche@gmail.com>
* Use pb.Reply instead of []byte with Reply.GetMessage() in llama grpc to get the proper usage data in reply streaming mode at the last [DONE] frame
* Fix 'hang' on empty message from the start
Seems like that empty message marker trick was unnecessary
---------
Co-authored-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
* Read jinja templates as fallback
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Move templating out of model loader
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Test TemplateMessages
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Set role and content from transformers
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Tests: be more flexible
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* More jinja
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Small refactoring and adaptations
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat(backend): add stablediffusion-ggml
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(ci): track stablediffusion-ggml
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fixups
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Use default scheduler and sampler if not specified
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fixups
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Move cfg scale out of diffusers block
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Make it working
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix: set free_params_immediately to false to call the model in sequence
https://github.com/leejet/stable-diffusion.cpp/issues/366
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* feat(bark-cpp): add new bark.cpp backend
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* build on linux only for now
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* track bark.cpp in CI bumps
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* Drop old entries from bumper
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* No need to test rwkv specifically, now part of llama.cpp
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
Use pb.Reply instead of []byte with Reply.GetMessage() in llama grpc to get the proper usage data in reply streaming mode at the last [DONE] frame
Co-authored-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
* feat(backends): Drop bert.cpp
use llama.cpp 3.2 as a drop-in replacement for bert.cpp
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore(tests): make test more robust
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* return 401 instead of 403, provide www-authenticate header, redirect to the login page, add cookie token support
* set cookies completely through js in auth page
* feat(ui): move model detailed info to a modal
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: add static asset
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* fix(diffusers): allow to specify width and height without enable-parameters
Let's simplify usage by not gating width and height by parameters
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
* chore: use sane defaults
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
---------
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
