Discover the Best AI Resources

SolarWM: Open Data and Scalable Training for Long-Horizon Video World Models

Provides an end-to-end, reproducible foundation for camera‑controllable, long‑horizon video world models — converting 1.43M clips from 10 datasets into a unified canonical corpus and releasing data, pipelines, recipes, and weights. Introduces backbone‑native adaptation and a three‑stage training recipe to produce 5B–33B models that enable minute‑to‑hour rollouts after training on 5s sequences.

Language Models Can Control Their Own Attention

Introduces Declarative Attention (DA), a zero-shot protocol that has LMs declare which parts of long context to attend to during chain-of-thought, letting the runtime build dynamic attention masks and skip most KV-cache reads. Produces large token savings (up to ~52% on Gemma-4-31B) with modest accuracy loss.

Iris: Climbing to the Search Frontier

Presents two LLM-based search agents (Iris-mini and Iris-pro) trained by alternating supervised fine-tuning and reinforcement learning against live web search. Key features: web-graph-derived multi-hop tasks with entity abstraction, SFT–RL climbing, inference-time context management, and state-of-the-art open-source benchmark results.

Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments

Reconstructs executable terminal workspaces from recorded agent trajectories and synthesizes verifiable single- and multi-round coding tasks for agent training; it replays file operations, uses an LLM completion agent to fill missing files/dependencies, and verifies tasks with autogenerated test suites.

Unlocking Lossless Speedups in LLMs via Discrete Diffusion

Uses lightweight discrete-diffusion adapters to draft blocks of tokens in parallel while keeping the original autoregressive model as the final arbiter, enabling lossless speedups (up to 3×) and compatibility with existing AR weights via a Diffusion Distillation phase.

Rethinking On-Policy Distillation of Large Language Models II: One Training Example

Studies on-policy distillation (OPD) at the data-minimal limit by training on a single query, measuring state coverage and alignment dynamics, and showing OPD is often data-overfed but algorithm-starved.

Compile by Training: Turning Natural-Language Specifications into Local Neural Functions

Converts a natural-language function specification into a reusable local neural function by using teacher models to synthesize examples and finetuning a small adapter for a compact interpreter. Achieves higher semantic accuracy (83.6% on FuzzyBench-Hard) at the cost of roughly one minute compile time; produces versionable PAW artifacts for local deployment.

FlowBalance: Verifier-Grounded Self-Improvement from On-Policy Reasoning Experience

Combines sparse verifier outcomes with dense privileged‑hindsight token scoring to learn an outcome‑calibrated, normalized distribution over complete responses for on‑policy self‑improvement. Key features: sign‑gated guidance (retain/reverse/disable per verifier advantage), profiled trajectory balance with one log‑partition per rollout group, and explicit correction against false‑positive self‑guidance.

Why Gated DeltaNet Survives 4-Bit Quantization: NVFP4 W4A4 for the Recurrent Half of a Hybrid 27B LLM

Shows that fully quantizing all 496 linear layers—including the recurrent Gated DeltaNet—of a hybrid 27B LLM to 4-bit NVFP4 W4A4 preserves benchmark accuracy while reducing model size to 17.5 GiB and improving prefill speed; includes a calibrated NVFP4 checkpoint.

LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes

Builds high-fidelity image generators by pairing a 6B Diffusion Transformer with a frozen LLaDA2.0-Mini vision-language module, relying on extensive image-only pretraining and mid-training; model weights, training code, and recipes are released.

Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning

Compresses KV cache for long-chain reasoning by keeping prompt tokens and evicting remaining entries uniformly at random per attention head; across four models and six reasoning tasks it matches the strongest prior evictor while delivering 32–43% higher vLLM throughput. Relies on prompt protection and redundancy across heads/text to retain reasoning traces; suitable when static memory budgets and higher serving throughput are priorities.

Qwopus3.8-27B-Flash-GGUF

Fine-tuned variant of Qwen3.8-27B optimized to reduce reasoning cost and wall-clock latency for long-running agent workloads. Delivers ~12.8% faster decoding and higher MTP draft acceptance while cutting runaway reasoning, at a small MMLU-Pro accuracy trade-off (−1.45 pp).