Commit Graph

569 Commits

Author SHA1 Message Date
Daniel Garvey
f206ecc635 reenable compilation in vicuna pipeline, add flags (#1505)
* replace vicuna.py backend with pipeline

* add some memory management to fist vicuna compile

reenable compilation
2023-06-07 09:49:27 -07:00
Stefan Kapusniak
a187e05ae6 Prevent having no cuda devices breaking the UI (#1503)
Don't break the UI when the LLM tab only wants cuda devices but there
aren't any.
2023-06-06 11:41:16 -07:00
Gaurav Shukla
8c21960486 [vicuna] Set only cuda devices in vicuna UI for now
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-06-06 22:15:20 +05:30
Gaurav Shukla
be62fce676 [vicuna] Fix vicuna chatbot (#1499)
Signed-off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-06-06 09:23:32 -07:00
PhaneeshB
f23b778a6c remove old vicuna scripts 2023-06-06 21:35:58 +05:30
PhaneeshB
436edf900d add vic sharded pipeline 2023-06-06 21:35:58 +05:30
Gaurav Shukla
ed58c2553f [vicuna] Integrate vicuna in shark studio
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-06-06 20:57:48 +05:30
Stefan Kapusniak
f2ca58e844 Add .csv and .json param info to output gallery (#1495) 2023-06-06 07:08:34 -07:00
Ean Garvey
1dbcc736eb [SD] (RDNA2) Enable new tuning for sd1.4 (#1498) 2023-06-06 06:48:58 -07:00
Phaneesh Barwaria
a83808ddc5 Vicuna cuda on A100 40G (#1496)
* vic chat with memory management (precompiled vmfb)

* fix vmfb path and download
2023-06-06 15:10:33 +05:30
Ean Garvey
d0ba3ef8fa disable use_tuned on SD1.4 for rdna2 (#1490)
this is a temporary measure while we retune SD1.4 for rdna2. The current config fails during iree-compile.
2023-06-05 19:46:16 -05:00
Stefan Kapusniak
8400529c2c Fix output gallery not using shark_tmp (#1493)
This fix the gallery component of the  output gallery dumping temporary
files into the standard folders rather than shark_tmp so those files never
got cleared out on restart and would build up.
2023-06-05 16:23:49 -05:00
Ean Garvey
6296ea4be9 fix config handling for sd1.4 on rdna2 (#1489) 2023-06-05 00:02:30 -07:00
Ean Garvey
e677178bcc Replace RDNA2 SD lowering configs. (#1486) 2023-06-05 00:57:43 -05:00
Ean Garvey
218ed78c40 Change instances of input_type='mhlo' to 'auto' (#1482) 2023-06-02 16:43:47 -05:00
Stefan Kapusniak
6046f36ab6 UI/Web: Fix upscaler stop button (mostly) (#1479)
* UI/Web: Fix upscaler stop button

* Hook the cancel_sd function up to the Stop button.
* Adds checks for SD_STATE_CANCEL in the upscaler ui inference function.
* Set and check for SD_STATE_IDLE, SD_STATE_CANCEL in the upscaler
pipeline.

* UI/Web: lint fixes for upscaler stop button fix

---------

Co-authored-by: powderluv <powderluv@users.noreply.github.com>
2023-06-01 22:26:55 -07:00
Phaneesh Barwaria
f0a4e59758 LLM Pipeline Wrapper (#1477)
* [LLM] Add LLM pipeline

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>

* add base pipeline and stableLM

* StableLM on UI - full block

* add SLM default model name

* add vicuna with pipeline

* add one token gen api for vic

* Fix stableLM bugs

* debug vic memory

* lint fix

---------

Signed-off-by: Gaurav Shukla <gaurav@nod-labs.com>
Co-authored-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-05-31 10:17:20 -07:00
Stefan Kapusniak
1ddef26af5 Web/UI: Add an Output Gallery tab for SD (#1470)
* WebUI: Adds an Output Gallery tab

Adds an new Output Gallery tab to the ui/webui with these features:

* Subdirectory select dropdown listing subdirectories at any depth below
the <output_dir>/generated_imgs directory,
* Large, full height, gallery area displaying the images in the selected
subdirectory. Shows nod logo when no images are in the selected
subdirectory.
* Slider that changes the number of columns of images that the gallery
displays from between 1 to 16 columns (defaults to 4).
* Expandable parameter info panel showing any generation parameters
saved in the file of the selected image for PNGs, alternatively the
image's EXIF data for JPEGs
* Send to buttons for txt2img, img2img, inpaint, outpaint and upscaler.
* Auto update of gallery and gallery label (to show generation status),
when a new image is generated by any of the stable diffusion tabs, and
is outputted to the currently selected subdirectory.
* Command line option for enabling and disabling the output gallery
(defaults to enabled)
* Command line option for following symlinks when getting entries
for the subdirectory list (defaults to off, as Python os.walk doesn't
check for circular references if following symlinks)

* Reformat with black

Reformat changes with black and then adjust some places where black's
formatting then needed some rephrasing of the code to make things
clearer.

* Add back transformers and sd_cancel imports

Adds back the transformers import in index.py needed for .exe
generation. Add comment so it doesn't get mistakenly removed
next time.
Adds back sd_cancel import in upscaler.py that is currently unused
but should be being used for the 'Stop' button.
2023-05-30 13:47:48 -07:00
Ean Garvey
1b4f4f5f4d Fix download path for SD1.4 Unet. (#1469) 2023-05-26 11:59:51 -07:00
Elias Joseph
73cd7e8320 added full vicuna to vicuna.py 2023-05-26 22:06:40 +05:30
Ean Garvey
19c0ae3702 Cleanup SD pipeline utils (#1466) 2023-05-25 12:50:11 -05:00
Ean Garvey
54e57f7771 Revive SD downloads from shark_tank. (#1465) 2023-05-25 12:03:21 -05:00
PhaneeshB
6d64b8e273 vic and slm common generation base 2023-05-25 20:29:41 +05:30
PhaneeshB
a8ea0326f5 correct SLM saved vmfb naming 2023-05-25 20:29:41 +05:30
PhaneeshB
58e9194553 add Lists import 2023-05-25 20:29:41 +05:30
PhaneeshB
eb360e255d remove unused imports 2023-05-25 20:29:41 +05:30
PhaneeshB
a6f88d7f72 refactor mlir compile 2023-05-25 20:29:41 +05:30
Phaneesh Barwaria
f5ce121988 SLM on Sharkstudio (#1454)
* localize import, fix file reading, device cpu

* extract out model args
2023-05-19 11:21:08 -07:00
Ean Garvey
991f144598 Add iree hidden imports to SD spec (#1456)
* Add iree hidden imports to SD spec

* Update shark_sd_cli.spec
2023-05-19 11:19:16 -07:00
PhaneeshB
09bea17e59 fix #2 SLM in SharkStudio 2023-05-18 00:56:22 +05:30
Daniel Garvey
aefcf80b48 swap to cpu an remove hardcoded paths (#1448)
Co-authored-by: powderluv <powderluv@users.noreply.github.com>
2023-05-17 10:53:34 -07:00
PhaneeshB
512235892e fix SLM for SharkStudio 2023-05-17 22:34:30 +05:30
PhaneeshB
6602a2f5ba add continuous output for CLI 2023-05-17 18:33:46 +05:30
powderluv
8ee2ac89f8 Rename sharded_vicuna_fp32_web.py to vicuna_web.py 2023-05-16 09:41:35 -07:00
powderluv
60cb48be2e Rename sharded_vicuna_fp32.py to vicuna.py 2023-05-16 09:40:51 -07:00
powderluv
86a215b063 Delete sharded_vicunia.py 2023-05-16 09:37:39 -07:00
powderluv
d6e3a9a236 Delete standalone_vicuna.py 2023-05-16 09:37:26 -07:00
Daniel Garvey
4731c1a835 prevent loading tokenizer on import (#1432)
also adds sentencepiece dep for exe
moved vicuna imports to after an if statement
in general we should avoid importing files that load whole models as
global variables
2023-05-12 19:11:45 -07:00
Gaurav Shukla
e0cc2871bb [SD] Yield 2 tokens at a time in vicuna
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-05-11 23:49:01 +05:30
Gaurav Shukla
649f39408b [SD] Fix vicuna response
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-05-11 18:06:21 +05:30
Gaurav Shukla
9e07360b00 [SD] Standalone vicuna with web
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-05-11 17:23:44 +05:30
Gaurav Shukla
7b74c86e42 [SD] Fix SAMPLE_INPUT_LEN import issue
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2023-05-11 15:41:43 +05:30
Eliasj42
fa833f8366 fixed spacing issue with chat-bot (#1417)
Co-authored-by: Elias Joseph <elias@nod-labs.com>
2023-05-10 16:07:50 -07:00
Gaurav Shukla
fcb059aa38 [SD] Integrate vicuna in the web (#1410) 2023-05-10 11:30:22 -07:00
PhaneeshB
517c670f82 vicuna chat cli 2023-05-10 22:55:06 +05:30
Eliasj42
59df14f18b added vicuna demo (#1408)
Co-authored-by: Elias Joseph <elias@nod-labs.com>
2023-05-09 21:18:20 -07:00
Daniel Garvey
7a4a51ae73 vulkan vic f16 (#1404)
Co-authored-by: dan <dan@nod-labs.com>
2023-05-08 16:46:53 -07:00
Eliasj42
54ce3d48ca added standalone vicuna script (#1401)
Co-authored-by: Elias Joseph <elias@nod-labs.com>
2023-05-05 18:05:52 -05:00
Daniel Garvey
6ca1298675 maximizes window size for webview launch (#1394) 2023-05-04 20:43:06 -07:00
jinchen62
bbef7a6464 Redesign model manager webui (#1391) 2023-05-04 20:41:29 -07:00