Mark Space offline pending architecture reassessment

tune: cosine floors -0.05 (0.55→0.50 high, 0.50→0.45 moderate) for 4K pool

tune+ux: z-thresh 1.8 + new examples + 'development practices' copy (README.md)

tune+ux: z-thresh 1.8 + new examples + 'development practices' copy (app.py)

tune+ux: z-thresh 1.8 + new examples + 'development practices' copy (confidence.py)

tune: high tier cosine floor 0.60 → 0.55 (promote in-distribution wins)

docs: add 'About the model' section + calibration update + fix v1.4→v1.2 refs

fix: force SentenceTransformer to CPU (ZeroGPU)

fix: import spaces before torch-loading modules (ZeroGPU)

config: drop undocumented hardware: yaml field

ui: sanitize_html=False on CTA-rendering Markdown components

ui: fix CTA visibility in dark mode (explicit-color HTML)

copy: emphasize public repos, drop PM tagline + vortex example

rename: Feature Finder (app.py)

rename: Feature Finder (README.md)

ux: conversion-oriented CTA + diversified examples (confidence.py)

ux: conversion-oriented CTA + diversified examples (app.py)

ux: bump PR body max_new_tokens 350→600 so completions don't truncate

ux: accept github.com/repo + owner/repo + SSH + with-subpath URL variants

fix: unpack BatchEncoding into generate() for transformers 5.x

fix: drop token kwarg from load_adapter; add startup diagnostics

fix: use base.load_adapter() instead of PeftModel wrap (transformers 5.x integration)

fix: add missing 'import os' for USE_SPACES_GPU read

H200 support: conditional @spaces.GPU decorator + hardware update (app.py)

Switch to persistent H200 (matches the Space's granted hardware)

Switch hardware to zero-h200 to match the Space's existing Zero-GPU grant

Replace with MHPD paper recommender (v1.2 backend) to reuse Zero-GPU allocation

update system prompt

update system prompt

update app

bump version

update inputs

update deps

update deps

zerogpu updates

add torchvision

add SpaceThinker

fix example imgs

add spacy download

add spacy download

adding install in app

adding install in app

Merge branch 'main' of https://huggingface.co/spaces/remyxai/VQASynth into main

update reqs

Update README.md

pin all deps

fix tqdm

specify transformers version

bump numpy

update reqs