TPIPS — Early Fusion (R1/R2 register tokens)

Text-conditioned perceptual image similarity, built on Qwen/Qwen3-VL-Embedding-8B. This repo holds the early_fusion checkpoint (one of three TPIPS models, each in its own repo — see the table at the bottom). Code and full docs: https://github.com/adobe-research/TPIPS.

Early fusion. (img0, img1, text) are encoded jointly with two register tokens R1 / R2 under a block-sparse attention mask. The pairwise score is cos(R1, R2) (symmetrized over both orderings) (higher = more similar). Odd-one-out probabilities are a softmax over the three "other-pair" scores divided by the temperature; 2AFC compares the two reference-candidate scores. The pairwise score is the model's raw output (temperature is applied at the probability step).

Property	Value
Base model	Qwen/Qwen3-VL-Embedding-8B
Pairwise score	`cos(R1, R2)` (symmetrized over both orderings)
Fine-tuning	LoRA (r=64, α=128) on the LLM layers; vision tower frozen
Register tokens	2 (R1 sees text+img0; R2 sees text+img1)
Temperature	0.05 (applied at the probability step)
Prompt `X`	`Measure the similarity between the two images based on X.`

Usage

TPIPS supports Python 3.10 and later. Install matching PyTorch and torchvision builds from the official PyTorch installer, then install TPIPS:

pip install tpips

import tpips
from PIL import Image

model = tpips.load_model("early_fusion", device="cuda")
a = Image.open("a.jpg").convert("RGB")
b = Image.open("b.jpg").convert("RGB")

similarity = model.similarity(a, b, factor="lighting")  # higher is more similar
distance = model.distance(a, b, factor="lighting")      # lower is more similar

The first call downloads the selected TPIPS checkpoint and its Qwen backbone. A CUDA GPU is recommended; FlashAttention is optional.

The TPIPS models

Model	Repo
Embedding (late fusion)	`sywang/TPIPS-Embed-Qwen3VL-8B`
Early Fusion	`sywang/TPIPS-EarlyFusion-Qwen3VL-8B`
Activation Distance	`sywang/TPIPS-ActDiff-Qwen3VL-8B`