pharmapsychotic
ce9d271aa1
Support for many different caption models:
...
blip-base, blip-large, blip2-2.7b, blip2-flan-t5-xl, git-large-coco
2 years ago
pharmapsychotic
9204a33786
Fix to handle updated Colab runtime
2 years ago
pharmapsychotic
42b3cf4d9e
Bunch of updates! ( #40 )
...
- auto download the cache files from huggingface
- experimental negative prompt mode
- slight quality and performance improvement to best mode
- analyze tab in Colab and run_gradio to get table of ranked terms
2 years ago
pharmapsychotic
a8ecf52a38
.
2 years ago
pharmapsychotic
1ec6cd9d45
Update to nicer BLIP packaging
2 years ago
pharmapsychotic
180cbc4f7b
Remove libraries from pip install
...
This fixes issue on Colab with conflicting numpy versions and exception importing transformers
2 years ago
pharmapsychotic
8b689592aa
Free Colab is too much of a potato to handle model swapping so just pick it in the setup cell.
2 years ago
pharmapsychotic
2ffcd80b4e
Update huggingface links in notebook
2 years ago
pharmapsychotic
884aab1a26
Default to ViT-L, lower intermediate count for Colab with ViT-H
2 years ago
pharmapsychotic
abbb326f93
.
2 years ago
pharmapsychotic
979fca878e
Download preprocessed cache files for ViT-L
2 years ago
pharmapsychotic
f22be02819
Prep merge to main
2 years ago
pharmapsychotic
19586d3d0d
Update description with link to SD 2.0 announcement post
2 years ago
pharmapsychotic
efee3fe0d7
Tweak defaults so runtime acceptable on T4 GPU in Colab
2 years ago
pharmapsychotic
ad01cadbef
Config changes to stay in 16GB VRAM available on Colab
2 years ago
pharmapsychotic
8d2de646b6
Quick note on Colab compatibility before I head out for turkey
2 years ago
pharmapsychotic
429d490901
First test version with OpenCLIP and ViTH!
2 years ago
pharmapsychotic
55c922a48a
Update notebook batch processing with option to rename files so can be used with [filewords] in Dreambooth!
...
- new `quiet` config option so CLIP Interrogator doesn't print and tqdm
- `max_flavors` option to each interrogate method
2 years ago
pharmapsychotic
6953206901
.
2 years ago
pharmapsychotic
31b1d22e82
Batch processing in notebook and run_cli!
2 years ago
pharmapsychotic
6139576f88
Use Gradio UI for the notebook
2 years ago
pharmapsychotic
9ce0f68ab3
CLIPInterrogator -> Interrogator
2 years ago
pharmapsychotic
7a2ac9aa57
Add to pip!
2 years ago
pharmapsychotic
27b9915dfa
Update notebook to use interrogator as library
2 years ago
pharmapsychotic
b423f7b090
Fix for when uploaded image has alpha channel
2 years ago
pharmapsychotic
2486589f24
Include example output
2 years ago
pharmapsychotic
6a0f6d457d
Version 2! Flavors upgraded from 400 to 100,000! Tests for ordering of common pieces. Chaining of flavors to progressively improve prompt. ipywidgets ui for easy uploading and interrogation. Other stuff I forget by now.
2 years ago
pharmapsychotic
b8362c596c
Fix the CLIP model notes and add links
2 years ago
pharmapsychotic
63e448ef81
Fix notebook problem
2 years ago
altryne
405a735bc5
Note about stable diffustion
...
Per twitter comment, this is the proper way to interrogate stable diffusion generated images.
2 years ago
pharmapsychotic
f36e13caef
Change load_list calls to directly use path so users can experiment with custom list with one line change
2 years ago
pharmapsychotic
2b4efdc76b
Some iteration and improvement
2 years ago
pharmapsychotic
1080560c48
Ported my CLIP Interrogator over to colab to share
...
The CLIP Interrogator uses the OpenAI CLIP models to test a given image against a variety of artists, mediums, and styles to study how the different models see the content of the image. It also combines the results with BLIP caption to suggest a text prompt to create more images similar to what was given.
2 years ago