An uncensored NVFP4-quantized text encoder for MiniMax-H3 video generation that fits on a single 16 GB GPU. Mixed-precision bake (mostly NVFP4, embedding left as INT8), preserves ConvRot rotation semantics, and includes the unrotate step required to avoid corrupted conditioning.
Generates short videos with stereo audio from text prompts using a MiniMax‑H3 checkpoint; community‑uploaded on Hugging Face and distributed under Apache‑2.0. Tuned toward stylized creature and floral visuals and updated frequently per the model card.
A 27B-parameter causal language model with a native vision encoder for image/video+text understanding, long-horizon agentic tasks, and tunable thinking-mode reasoning. Native 262,144-token context (extensible to 1,000,000) and production-focused inference recipes.
A LoRA adapter for MiniMax-H3 that enables joint video + synchronized stereo audio generation in as few as 4 sampler steps, cutting sampling time roughly ~5×; early prototype under-trained, so 6–8 steps or newer checkpoints give better sharpness.
Provides ComfyUI-compatible pruned/curve-form LoRA conversions of the MiniMax‑H3 Turbo 4-step audio‑video generation preview, including further-trained ckpt500 EMA and non‑EMA variants and an example ComfyUI workflow for low-step experiments.
Provides 1,080,814 images extracted from ~65,000 digitised British Library book volumes (c.1510–c.1900), split into four algorithmic image-type configs and packaged as parquet for image–text multimodal research and retrieval.
Evaluates schema-guided structured extraction from documents: given a document and a JSON schema, systems must return a schema-valid JSON with page-and-box grounding. Covers 370 documents (4,869 pages) across 8 business domains and 67 document types; scores value accuracy, word/page grounding, and long-list completeness.
Provides per-decision training samples for RL-driven command-line LLM agents: each record pairs a task prompt plus terminal history with a teacher's next-action in Terminus-2 JSON. Around 31k verifier-passing samples from 630 ATCB tasks, formatted for NeMo Gym's terminus_judge and licensed CC-BY-4.0.
Provides ~39 TB of pre‑beamformed (channel capture) ultrasound RF data and metadata in zea/HDF5 format for reconstruction, flow, and inverse‑problem tasks. Released under CC‑BY‑4.0 and curated for training and evaluating ultrasound/RF foundation models.
Turns a short prompt plus aspect ratio and duration into a structured, shot-by-shot audio-video description for text-to-audio-video generation. A PEFT LoRA on Qwen3.6-27B that expands timing, camera motion, continuity, and synchronized diegetic/non‑diegetic sound; text-only and requires MiniMax-H3 + LightX2V to produce final AV.
10,000-hour head-and-wrist egocentric dataset pairing synchronized head and wrist video with left/right 3D hand pose and optional full-body pose; provided in LeRobot/MCAP formats with episode-level semantic annotations and automated de-identification.
Packaged diffusers checkpoint of MiniMax H3 for image/text-to-short-video generation with native stereo audio; provided for direct use in diffusers image-to-video pipelines and aimed at easy integration into prototyping and production workflows.