You need to agree to share your Hugging Face information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

This adapter is a Model Derivative of MiniMax H3 and is governed by the MiniMax H3 Community License Agreement, including its Applicable Territory and Acceptable Use Policy. Request access only if the license allows your use.

Log in or Sign Up to review the conditions and access this model content.

ReDiverse: Restoring Diversity in Distilled Image and Video Generators by Probing Activations

Furkan Horoz1, Ozgur Kara1, Kaan Oktay2, Adil Kaan Akan2, Atakan Kara2, Anand Bhattad3,†, James Matthew Rehg1,†
1University of Illinois at Urbana-Champaign    2fal    3Johns Hopkins University
†Equal advising.

Project page arXiv 2611.18550 Code on GitHub All ReDiverse models

ReDiverse-FT adapter for Pika H3 Flex. Pika H3 Flex is a fine-tuned MiniMax-H3 that generates video with synchronized audio in six sampling steps. Like the other few-step generators, it gives near-identical clips across seeds. This rank-32 LoRA on the blocks up to transformer_blocks.45 is trained, with the recipe of the MiniMax-H3 adapter, to match the teacher's activations there, and is active in the first sampling step only; the remaining five steps run Pika H3 Flex unchanged.

Teacher 6-step student This adapter
MiniMax-H3 Pika H3 Flex, its text-to-video (FL2VA) LoRA of the teacher ReDiverse-FT LoRA, stacked on Pika's LoRA, rediverse_ft.safetensors (612 MB)

Pika H3 Flex, Iron Man flying in the sky: four seed-matched clips without and with ReDiverse

More Pika H3 Flex results are on the project page (Video Results, Pika H3 Flex tab).

Usage

Install the Pika H3 Flex package first (Install); it downloads the MiniMax-H3 base files and Pika's adapter on first use.

hf download ReDiverse/ReDiverse-Pika-H3-Flex --local-dir ReDiverse-Pika-H3-Flex
python ReDiverse-Pika-H3-Flex/generate_pika.py --prompt "a corgi running on the beach at sunset" --seeds 0-3 --out out --compare

This writes out/rediverse_ft/<seed>.mp4 and, with --compare, a seed-matched out/pika/<seed>.mp4 rendered by Pika H3 Flex alone from the same noise. The default is 768p, 16:9, 5 s (--resolution 480p, --aspect, --duration). The adapter was trained and tested for text-to-video.

License

A Model Derivative of MiniMax H3, released for research purposes and governed by the MiniMax H3 Community License Agreement, including its Applicable Territory (which excludes the European Union, the United Kingdom, the Republic of Korea and the United States of America) and Acceptable Use Policy. Check that the license allows your use before downloading or running Pika H3 Flex, MiniMax-H3 or this adapter. NOTICE holds the required notice. Powered by MiniMax H3.

Citation

@article{horoz2026rediverse,
  title   = {Restoring Diversity in Distilled Image and Video Generators by Probing Activations},
  author  = {Horoz, Furkan and Kara, Ozgur and Oktay, Kaan and Akan, Adil Kaan and Kara, Atakan
             and Bhattad, Anand and Rehg, James Matthew},
  journal = {arXiv preprint arXiv:2611.18550},
  year    = {2026}
}
Downloads last month
20
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for ReDiverse/ReDiverse-Pika-H3-Flex

Adapter
(3)
this model

Collection including ReDiverse/ReDiverse-Pika-H3-Flex

Paper for ReDiverse/ReDiverse-Pika-H3-Flex