LoRA adapters for Qwen/Qwen3.8-27B trained with supervised fine-tuning (SFT) on visual questions generated by PixelProof, as described in the paper PixelProof: Visual Question Generation through Forward–Inverse Agreement. In PixelProof, a coding agent writes a forward program that draws images and computes their answers and an inverse program that must recover each answer from the image alone; a question is kept only if the two agree. Each adapter is stored in its own subfolder. The repository contains no merged weights. Training uses transformers.Trainer and PEFT LoRA adapters on the language model's attention and feed-forward projections; the vision tower is not directly adapted. The main…
Independent publisher
Project PixelProof
PixelProof
Models in Library3
Datasets in Library1
Models on Hugging Face3
Followers—
Compare the published evidence
This profile connects the 4 model and dataset records shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Published tasks: Image and text to text, Visual question answering.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.
Models
LoRA adapters for google/gemma-4-31B-it trained with supervised fine-tuning (SFT) on visual questions generated by PixelProof, as described in the paper PixelProof: Visual Question Generation through Forward–Inverse Agreement. In PixelProof, a coding agent writes a forward program that draws images and computes their answers and an inverse program that must recover each answer from the image alone; a question is kept only if the two agree. Each adapter is stored in its own subfolder. The repository contains no merged weights. Training uses transformers.Trainer and PEFT LoRA adapters on the language model's attention and feed-forward projections; the vision tower is not directly adapted. The…
LoRA adapters for google/gemma-4-12B-it trained with supervised fine-tuning (SFT) on visual questions generated by PixelProof, as described in the paper PixelProof: Visual Question Generation through Forward–Inverse Agreement. In PixelProof, a coding agent writes a forward program that draws images and computes their answers and an inverse program that must recover each answer from the image alone; a question is kept only if the two agree. Each adapter is stored in its own subfolder. The repository contains no merged weights. Training uses transformers.Trainer and PEFT LoRA adapters on the language model's attention and feed-forward projections; the vision tower is not directly adapted. The…
Datasets
The frozen evaluation sets for the PixelProof paper. Each eligible replay-verified world contributes exactly five deterministic, pixel-distinct instances. The five instances of a world share its selected question text while varying the rendered scene and the gold answer, which the world's inverse program recovers from the image alone. The config names, the section column, and the ledgers/section files are holds spatially-steered worlds. The exclusions are the worlds that passed the 200-scene replay (1,301 profile-steered and 875 spatially-steered) but produced fewer than five pixel-distinct scenes; each one is documented in the corresponding ledgers/sectionexclusions.csv file. Each row…