Skip to main content
Search

# Image Generation

16 related articles

Subscribe to updates
  1. Low-rank adaptation learns one matrix: SingLoRA beats LoRA and LoRA+ on LLaMA-7B at 60% of their parameter budget

    Low-rank adaptation (LoRA) writes the weight update as a product of two small matrices, but their norms often differ sharply during training, producing imbalanced gradient flow and unstable optimization. SingLoRA instead writes the update as a single low-rank matrix times its own transpose, removing inter-matrix scale conflicts by construction and roughly halving the trainable parameters. Fine-tuning LLaMA-7B on MNLI, SingLoRA reaches 91.3% accuracy against 89.1% for LoRA and 90.2% for LoRA+, while using only 60% of their parameter budget.

  2. Vercel News

    Nano Banana 2.1 lands on AI Gateway: product recontextualization, mask-based edits and factuality all improve

    Google's Nano Banana 2.1 is now available on AI Gateway as google/gemini-nano-banana-2.1, so developers can call it directly. Compared with earlier Nano Banana models, it improves image generation and editing, covering product recontextualization, mask- and ink-based editing, and factuality, while maintaining Flash-tier latency and cost. It can be used through the AI SDK, the OpenAI-compatible Chat Completions API, or AI CLI.

  3. arXiv

    Pixel-space generators stop letting detail inflate global attention: low-frequency carriers cut ImageNet-512 FID by 2.84

    Pixel-space generators feed the full image to a Transformer as patches, so finer detail lengthens the global sequence and raises quadratic attention cost, tying detail to communication. SpecSlot expands each spatial anchor into slots, gives every slot a fixed disjoint group of DCT coefficients, and sends only the lowest-frequency slot per anchor into the global stage while all slots mix locally. At ImageNet-512, the W256/K4 configuration lowers FID by 2.84 over the FLOPs-matched JiT control with 3.9% fewer GFLOPs, and trails JiT-B/16 by 0.69 FID at 38.2% lower cost.

  4. arXiv

    Image-generation agents stop merely retrying: failure experience passes a counterfactual check before entering the skill library, reaching 86.1 on GenEval2

    Agentic image-generation pipelines have largely been read-only generate-verify-rewrite loops that discard repair experience or admit ungated updates into a skill library. VEvo-Harness expands each failed verdict into a hierarchical verification tree to locate root causes, then admits only causally valid skill patches through counterfactual dual-arm image replay, with memory organized into working, experience and pattern tiers. On the Qwen-Image-2512 backend it reaches 86.1 on GenEval2, 14.5 points above GenRouter, while continuing to evolve.

  5. Trainium stops demanding compiler-style code: a native PyTorch backend runs 89% of top text-generation architectures

    Running Hugging Face models on AWS Trainium now rests on a different foundation: TorchNeuron, a native PyTorch backend, exposes Neuron chips as an ordinary device with eager execution, torch.compile and native distributed APIs, replacing the PyTorch XLA-based optimum-neuron. By Hugging Face's comparison, out-of-the-box support among the 100 most-used text-generation architectures rises from 46% with Optimum Neuron v0.4.5 to 89%. The backend enters public beta on October 20, 2026.

  6. arXiv

    Diffusion sampling gets a dedicated corrector: FID drops from 9.58 to 2.07 on ImageNet, and unguided generation approaches CFG

    Diffusion sampling accumulates the model's own prediction errors, and a corrector trained on the base model's predictions against the same clean target lowers DiT-XL/2 FID on ImageNet class-conditional generation from 9.58 to 2.07, approaching CFG without guidance.

  7. Vercel News

    Microsoft's image models land on AI Gateway: one key for text-to-image, and five reference images for product and layout edits

    Microsoft's MAI-Image-2.6 and MAI-Image-2.6-Flash are now available on AI Gateway, letting developers generate and edit images through the AI SDK or AI CLI with one API key and no provider account. Both models support photorealistic imagery, text rendering, and targeted edits to products, portraits, and layouts, accept up to five reference images to guide edits or combine subjects, and output up to 1536 × 1536. MAI-Image-2.6 targets final assets where quality matters more than speed, while Flash targets faster, lower-cost generation for many iterations.

  8. arXiv

    Removing residual connections in the DDT condition encoder and fusing multi-depth features lets DDT-RFE improve four visual understanding tasks and ImageNet FID with fewer encoder blocks

    The authors remove the residual connections around Self-Attention and MLP in the condition encoder of the Decoupled Diffusion Transformer (DDT), stabilize training with zero-initialized output gates and a mixed orthogonal attention initialization, and form the encoder output by aggregating the input patch embedding with the outputs of all encoder blocks; under ImageNet-1k training at the stated resolution, patch size 2, 80 epochs and batch size 1024, DDT-RFE improves overall over DDT on image classification, semantic segmentation, object discovery and semantic correspondence while using fewer encoder blocks (12 vs. 20 at L, 14 vs. 22 at XL), and lowers ImageNet generation FID from 8.98 to 7.68 at L and from 7.27 to 6.81 at XL.

  9. arXiv

    LiFT lets looped DiTs keep improving beyond training depth: L/2 and XL/2 beat larger dense DiTs with fewer parameters and less inference compute

    LiFT supervises each loop of a recurrent DiT toward a point on a straight path from the model's initial velocity estimate to the flow-matching target, so a trained checkpoint can run far more inference loops than its training depth; on class-conditional ImageNet, L/2 and XL/2 checkpoints that reallocate inference compute toward recurrence reach lower FID than larger dense DiTs with fewer parameters and less inference compute, while at B/2 the dense model remains stronger.

  10. arXiv

    FaV-A generates positive-negative space compositions via a staged multimodal agent, with experiments and ablations suggesting gains over direct zero-shot MLLM baselines

    The work presents Form and Void Agent (FaV-A), a multimodal agent for staged positive-negative space generation: it first generates a base object, then analyzes its shape and spatial structure to identify candidate negative-space semantics, and finally produces compositional instructions for the final image generation stage; experiments and ablation analyses suggest FaV-A yields more visually coherent and semantically aligned positive-negative space compositions than direct zero-shot MLLM baselines.