Tag
Explore by tags
A PyTorch DTensor-native SPMD library for training and fine-tuning LLMs, VLMs, diffusion and retrieval models. Integrates with Hugging Face for day-0 model support, provides YAML-driven recipes, DTensor/FSDP2 parallelism and NVIDIA-optimized kernels (Transformer Engine, DeepEP, FlexAttn).
Generates and edits full-length music locally using a hybrid language-model planner and a diffusion transformer. Provides text- and audio-driven song synthesis plus selective editing (cover generation, repainting, vocal→BGM), metadata control and lightweight LoRA personalization, and runs on consumer GPUs/CPUs.
Provides a curated collection of hands-on tutorials, workflows and auxiliary files for training and using generative-model tooling (Stable Diffusion, Flux, WAN). Key items include a WAN 2.1 LoRA training tutorial and an articles collection covering DreamBooth, LoRA, LyCORIS and SDXL.
Provides a one-command CLI to fine-tune and post-train LLMs, with layer streaming that lets an 8B model be fine-tuned on a 4 GB laptop GPU. Auto-configures quantization, LoRA adapters, batching and evaluation gates, and supports export and serving workflows.
Unifies multimodal understanding, reasoning, and image generation in a single end-to-end architecture using the NEO-unify paradigm. Models pixels and words jointly without a separate visual encoder, and provides interleaved image–text generation, infographic editing, and GGUF/low‑VRAM inference options.
Diffusion-based generative model for scene and video synthesis, providing full Diffusers checkpoints and scene LoRA for fast adaptation. Includes Stage‑1 nano (1.3B) and pro (5B) variants and modular transformer/VAE components.
Depth-conditioned LoRA for Krea‑2 that extracts a depth map from any input image and generates new images preserving the original 3D structure and composition while changing content and style. Single 862MB LoRA, works with Krea‑2‑Raw and Krea‑2‑Turbo.
Generates short videos that preserve a reference person's identity from a single reference image as a LoRA adapter for LTX-2. Uses overlap reference conditioning with TASS‑RoPE source-phase tagging and an ArcFace identity loss; runs in ComfyUI via BFS Nodes and supports a 4‑panel character‑sheet mode for clothing/body consistency.
Enhances KREA-2 Turbo image generations with an aesthetic LoRA trained on a curated 100-image dataset to add stronger composition, richer lighting, softer atmosphere and refined textures; trigger with --preview for art-directed, cinematic outputs in text-to-image pipelines.
Generates videos from text and image+text prompts using a 30B Mixture-of-Experts model tuned for embodied intelligence; includes a refiner and structured prompt rewriter, and supports diffusers/SGLang runtimes with multi-GPU inference.
A 9B-parameter Qwen3.5-based multimodal model tuned to preserve chain-of-thought reasoning while eliminating repetition loops; restores native multi-token prediction, supports 1,048,576-token context, and targets research/red-team use.