#marathonthegame #marathon average rook day off be like:

／ Campaign Now Live — Enter to win prizes! Day 8 ＼ Follow & RT to get iconic key art from past titles 🎁 How to Enter: 1. Follow @REBHPortal 2. Retweet this post by 4/6 at 10:59 PM(EDT) 3. Reply with the hashtag RE30th for a chance to win even more amazing prizes!

1/2 Meet Wan2.7-Video — The Comprehensive Model for Controllable Video Storytelling! From single clips to full-scale narrative direction, we’ve built more than just a generator. We’ve built a director’s suite: • Multimodal control over performance and style via text, image, audio, and video. • Character customization with up to 5 reference inputs and voice profiles. • Video editing with simple, intuitive instructions. • Full-stack creative toolkit: generation, editing, cloning, restyling, continuation, and more. • Sustained improvements in visual fidelity, motion stability, and prompt adherence.

／ Campaign Now Live — Enter to win prizes! Day 2 ＼ Follow & RT to get iconic key art from past titles 🎁 How to Enter: 1. Follow @REBHPortal 2. Retweet this post by 3/28 at 10:59 PM(EDT) 3. Reply with the hashtag RE30th for a chance to win even more amazing prizes!

I did 24 for this LoRA, but you can get by with less. 10 or so would probably work well.

🚀 FantasyWorld is now open source! Recently accepted to ICLR 2026 and ranking #1 on the WorldScore leaderboard, this framework offers a unified approach to video and 3D generation. ✨Key Technical Highlights: ✅Direct 3D Inference: Input an image and camera trajectory to get frame-consistent video, dense depth maps, and point clouds instantly. ✅Dual-Branch IRG Architecture: An "Imagination" branch handles visual appearance while a "Geometry" branch ensures spatial accuracy. ✅Wan 2.2 Integration: The released 14B models leverage the latest Wan-DiT architectures for superior temporal consistency and resolution. ✅Ready for Downstream: Perfect for embodied AI navigation, XR content, and autonomous driving simulation. 🤖 Models: modelscope.cn/models/amap_cv… modelscope.cn/models/amap_cv… 💻 GitHub: github.com/Fantasy-AMAP/fant… 📄 Paper: modelscope.cn/papers/2509.21…

Attention Sinks and Compression Valleys in LLMs are Two Sides of the Same Coin Paper: arxiv.org/abs/2510.06477

Oh I need this model in my OpenClaw! SoulX-Singer, a great zero-shot singing voice synthesis model has been released on @huggingface Apache 2 license as well! Check out the demo: huggingface.co/spaces/multim… And the model: huggingface.co/Soul-AILab/So…

SE-Bench Benchmarking Self-Evolution with Knowledge Internalization huggingface.co/papers/2602.0…

Ever wonder why Vision-Language Models get distracted by empty space and the bottom of your images? Researchers from Shanghai University and Nankai University introduce a new way to fix biased AI attention. They developed two simple debiasing tools that clean up attention noise. By removing the model's tendency to favor tokens just because of their position and ignoring useless padding, they make token pruning much more accurate and efficient. The method outperformed 7 state-of-the-art pruning techniques across 10 benchmarks, delivering major performance gains for both image and video tasks. Attention Debiasing for Token Pruning in Vision Language Models Paper: arxiv.org/abs/2508.17807 Code: github.com/intcomp/attention… Our report: mp.weixin.qq.com/s/neBdXnBdD… 📬 #PapersAccepted by Jiqizhixin

MiniCPM-o 4.5 brings true "Omni" capabilities (Vision Audio Text) into a single 9B end-to-end architecture (Qwen3-8B SigLip2 Whisper CosyVoice2).🚀 modelscope.cn/collections/Op… ✅ Vision SOTA: OpenCompass score of 77.6—surpassing GPT-4o and Gemini 2.0 Pro in VLM tasks with just 9B params. ✅ Real-time Full Duplex: It "sees," "hears," and "speaks" simultaneously. Supports active interaction (it can interrupt or initiate conversation based on live video). ✅ World-Class OCR: Beats DeepSeek-OCR 2 and GPT-4o in document parsing (OmniDocBench leader). ✅ Voice Cloning: Clone any voice with just a short reference clip—integrated directly into the multimodal flow. ✅Quantization: 16 types of Int4 and GGUF formats available.

🚀 Introducing GLM-4.7-Flash — the strongest 30B-class open model that balances performance & efficiency! ✅ Top scores on AIME, GPQA, SWE-bench, τ²-Bench & more ✅ Optimized for agentic coding, tool use, and reasoning ✅ Runs locally with vLLM / SGLang / Transformers Perfect for developers who want Claude-level coding power at a fraction of the cost 💡 Try it now: 🤖 modelscope.cn/models/ZhipuAI… 🔗 chat.z.ai 📄 z.ai/blog/glm-4.7 🐙 github.com/zai-org/GLM-4.5

ꜛꜛꜛꜛꜛ rt for 1000 years of good luck ꜛꜛꜛꜛꜛ

#REBHFun #ResidentEvilRequiem #Biohazard lfg we out here ready and waiting. February needs to hurry up.

Introducing FLUX.2 [klein]. Blazing fast. Beautiful. Generate stunning images in under a second while maintaining exceptional quality. Great for fast editing, changing styles, and developing ideas from 0 → 1. Available via API, or run it locally - Klein 4B under Apache 2.0, Klein 9B as open weights. Try it for free in our demo app (link in the thread).

🚀Meet Qwen-Image-2512-Fun-ControlNet-Union 🎨 ✅ Supports 6 control types: Canny, HED, Depth, Pose, MLSD, Scribble ✅ Works in inpainting mode too ✅ Multi-resolution conditioning → better generalization ✅ Fine-tune with control_context_scale (0.7–0.95) for stronger control & sharper details 💡 Pro tip: Pair with detailed prompts for max stability! 🤖 Model: modelscope.cn/models/PAI/Qwe…

These @iamspecialized brand mountain liner bib short plus SWAT for women is 75 to 79 % off in extra small, small and large on competitive cyclist right now down to 30 bucks from 119.99 #cycling #holidaygiftguide

Normally every Comfy edit needs a render job — so I built a realtime node for live image filtering. Stack nodes, adjust sliders instantly, and save progressive outputs without rerunning the workflow.