Encoder-decoder is back 😈!!! In DeepSeek V4.1 Flash the first 20 layers build the global KV that the next 20 read from, so prefill costs about half. Rolled out over the last 24h: throughput doubled, and we are approaching 1T tokens/day on OpenRouter. 30% off to celebrate. Enjoy. Cheapest on the market, as always.

"Sphere Encoder 2" This paper fixes why one-step autoencoder generation looks blurry. It trains the latent rotation all the way from an encoded image to the 90° equator region where random generation actually starts. Then instead of forcing those highly ambiguous latents to reconstruct the exact source image pixel-by-pixel, it uses semantic, latent-consistency, and feature-space score-matching losses so the decoder can generate a sharp plausible image rather than average many possibilities into blur. On ImageNet, it reaches FDr6 4.92 with just 4 steps and ~23x less sampling compute than PixNerd, though its gFID is worse.

Me: Claude we need an optical encoder circuit Claude: Sure thing! Lets build it. This was a dream several months ago, and now it's reality. Soon, anyone will be able to make anything.

A compute-based Vulkan APV encoder has just been merged into FFmpeg. It runs at 1200fps on a 6900XT for 1080p422 10-bit encoding. Meant for low-overhead high-quality recordings, or low-latency 4K streaming over a 1Gbps link.

Do robot VLAs really need a separate vision encoder? From Pixels to Actions: Towards Native Vision-Language-Action Policies NEO-VLA is a 2B encoder-free VLA built directly on pixels and words. 🧱 Highlight 1: a unified backbone with no encoder Instead of CLIP or SigLIP, a light convolutional patch embedder feeds image tokens into the same 40-layer decoder as the text. On LIBERO average with a frozen backbone, Qwen3-VL-2B scores 79.5 and NEO-VLA 90.9. 🔍 Highlight 2: the action expert reads intermediate layers A flow-matching expert attends to keys and values from middle layers of the backbone. With a frozen backbone, reading early layers is strongest (94.9); after fine-tuning, a layer mix reaches 95.3. But wiring the expert to an encoder-based model's intermediate features drops frozen performance from 79.5 to 64.9, so the pretraining seems to matter. 🧠 Highlight 3: co-training with understanding data Action-only fine-tuning gives 10.7% on RoboTwin 2.0 and loses the ability to output text. Mixing in understanding data reaches 69.4% and keeps VQA scores within about 1.5 points. LIBERO hits 97.4, but it's weak when objects or goals change (LIBERO-Pro: position 3.0, task 6.3), and the real-robot evaluation is small: 3 tasks, 20 trials each. #VLA# #Robotics#

TL;DR: One encoder and one JSON schema turn documents, images, audio and video into structured, grounded text. It posts the top reported score on document parsing. Youtu-Parsing-Omni: One Encoder, One Schema, All Modalities Key points ・🧩 A single shared Transformer encodes images, audio and video (initialized from Qwen3-0.6B-Base) ・⏱ Rotary positions split into time/height/width, so video and sound share one timeline ・📐 OmniSchema puts 10 tasks (documents, chemistry, sheet music, charts, audio, video...) into one JSON format ・🔁 Post-training combines OSAD distillation from a self-refreshing teacher with schema-aware RLVR ・📄 OmniDocBench v1.6: 96.96 overall, the highest reported ・🎬 OmniParsingBench: 75.08, best among open-weight models; text-rich video 75.62 vs 65.36 Gemini-3-Pro still scores higher at 77.44, and it trails Qwen3-Omni on Video-MME-v2. ASR is also a weak spot. Still, a promising step toward an "any-input parser" front end for LLM agents. #OmniModal# #DocumentParsing#

TL;DR Advanced encoders were losing to older models on sparse retrieval for one surprising reason: a vocabulary mismatch. Fixing it with just 20k adaptation steps sets a new BEIR state of the art. Title: Why Advanced Encoders Lag on Sparse Retrieval? The Answer and an Approach to Bridging Vocabulary Gaps URL: Key points 🔤 Raw case-sensitive vocabularies were splitting single meanings across redundant surface forms 🧩 The proposed Vocabulary Transfer method migrates vocabularies in 3 lightweight steps 📊 ModernBERT-VT hits nDCG@10 52.4 on BEIR, a new SOTA 💥 A collapsed RoBERTa-large jumps from BEIR score 1.4 to 51.3 ⚡ Uses under 0.2% of the original pretraining token budget 🧪 Near-optimal performance reached in just 500 MLM steps 🧬 Also works for domain-specific vocabularies like chemistry It's a nice reminder that what looked like an architectural ceiling turned out to be a fixable vocabulary design problem. #InformationRetrieval# #LLM#

Seeing InternVideo-Next explicitly characterize its architecture as "Encoder–Predictor–Decoder (EPD)" and defining the Predictor as a "latent world model" confirms a clear convergence in the field. 📉 From our early work on Context AutoEncoder (CAE, — where we decoupled the Encoder, Predictor ("Regressor" in the paper), and Decoder — to Meta's I-JEPA/V-JEPA shifting entirely to latent prediction, and now this. It seems we are all validating the same core intuition: Understanding ≠ Pixel Reconstruction. 💡 Decoupling representation learning from high-frequency detail generation to build World Models in latent space is evidently the path forward. Glad to see our early intuition resonating with the latest SOTA. 🫡 #AI# #ComputerVision# #WorldModel# #JEPA# #CAE#

Love the fact that Jev is bringing focus to structured / open-ended classification problems with encoder-style models. There is so much to be unlocked here... it brings a new dimension to LLM workflows with a lot of interesting opportunities. Interestingly enough, this was one of my earliest research interests. See the paper below where I trained a multimodal BERT model to solve >1000 multi-class classification problems with a single model. These types of models are really capable and can do so much if you train / scale them properly! However, it's worth mentioning that Jev goes well beyond basic encoder classification. Some examples of novel components are: - Generalizing to any classification space in a scalable manner with good inference efficiency (probably some improved version of CLIP-style contrastive approach). - In-depth, data-centric research to do this properly and maximize model performance with training scale. - Figuring out how to structure the model in a way that yields really impressive inference efficiency. These are all changes that create something completely different than just encoder + classification. Even small details make a massive difference in adoptability and utility for these types of models. Jev (or similar models) provide an orthogonal tool to standard autoregressive LLMs that is really helpful and can be integrated into LLM workflows with minimal added cost. There are so many directions we can go with these models IMO. Need to do a deeper dive and will follow-up with a full writeup soon.