Commit Graph

11 Commits

Author SHA1 Message Date
kingbri
1919bf7705 Launch: Make exllamav2 requirement more friendly
Add the ability to use an unsafe config flag if needed and migrate
the exl2 check to a different file within the exl2 backend code.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-02-02 23:36:17 -05:00
kingbri
b827bcbb44 Sampling: Cleanup and update
Cleanup how overrides are handled, class naming, and adopt exllamav2's
model class to enforce latest stable version methods rather than
adding multiple backwards compatability checks.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-02-02 23:36:17 -05:00
kingbri
d3781920b3 OAI: Split up utility functions
Just like types, put utility functions in their own separate module
based on the route.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-02-02 23:36:17 -05:00
Alexander Abushady
d7c18855e7 added quadratic sampling (#56)
* added quadratic sampling

* Update sample_preset.yml

* oops missed a spot

* Sampling: Fix smoothing factor semantics
2024-02-02 22:12:59 -05:00
kingbri
4a7b8b1b7a Samplers: Add dynamic temperature
Does not work if max_temp is less than or equal to min_temp. Sampler
validation will have to be refactored in the future, so the dynamic
temperature check will also be changed.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-31 01:20:59 -05:00
kingbri
751627e571 OAI: Add fasttensors to model load endpoint
Also fix logging when loading prompt templates.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-25 01:08:02 -05:00
kingbri
fc4570220c API + Model: Add new parameters and clean up documentation
The example JSON fields were changed because of the new sampler
default strategy. Fix these by manually changing the values.

Also add support for fasttensors and expose generate_window to
the API. It's recommended to not adjust generate_window as it's
dynamically scaled based on max_seq_len by default.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-25 00:15:40 -05:00
kingbri
90fb41a77a Model: Fix prompt template initialization
The previous commit iterated through multiple try conditions which
made it so the user has to provide a dummy prompt template. Now,
template loading is fallback based.

Run through a loop of functions and return if one of them succeeds.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-25 00:15:40 -05:00
kingbri
740b0215dd Model: Dynamically scale generate_window
Allows for adjustment of reservation space at the end of the context
before rolling it. This should be scaled as a model's max_seq_len
goes up.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-25 00:15:40 -05:00
kingbri
de0ba7214c API: Add template switching and unload endpoints
Templates can be switched and unloaded without reloading the entire
model.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-25 00:15:40 -05:00
kingbri
78f920eeda Tree: Refactor code organization
Move common functions into their own folder and refactor the backends
to use their own folder as well.

Also cleanup imports and alphabetize import statments themselves.

Finally, move colab and docker into their own folders as well.

Signed-off-by: kingbri <bdashore3@proton.me>
2024-01-25 00:15:40 -05:00