Add back mega pipeline.

Add missing auth token.

Remove mega custom pipeline to save memory, use euler-a scheduler with 20 steps to greatly speed up image generation.

Move pipeline back out for troubleshooting.

Fix deps.

Fix dep.

Add ftfy package.

Update gradio sdk version to 3.9.1, update other dependencies.

Reduce text and image size to reduce compute load.

Try inline pipeline to reduce memory usage.

Fix revision tag.

Use float32 for CPU compatibility.

Add auth token for diffusion model download.

Replace DALLE-mini with Stable Diffusion pipeline.

Fix formatting for "Melee or Ranged Weapon Attack"

Provide default description for image generation if missing, add repetition penalty of 1.2 to text generation

Add result example files, note how to call the API

Block execution if name is blank, hide HTML output

Update description, try update for API support

Remove extra css styling, hide HTML output, add type hints for run function

Filter extra 'Actions' text from result HTML, try two column layout, add labels to app components, rename from Test Space to MonsterGen

Automatically crop card png output

Lengthen output png, add additional formatting

Fix project config color, shorten output png

Reduce output png size, clean up template file, clear \r chars from text output

Fix README.md

Use python version 3.9.10, add logging

Update packages.txt to use chromium instead of chromium-browser

Add packages.txt with chromium-browser

Improve card formatting, render html into png for return

Add card formatting

Combine interfaces

Specify mega weights

Specify verbose logging, disable grads

Specify float16 to save memory

Specify mini weights to see if memory is an issue

Specify cpu device

Add debugging for image generation

Add image run code and update requirements.txt

Add model run code and requirements.txt

Upload export.pkl

Add application file

initial commit