AI Model2026
An uncensored GGUF build of GLM-5.2 that applies weight “abliteration” to remove refusal filters and produce a locally runnable text-generation model; includes quantization conversions and shard-merge instructions, intended for experimental research rather than production use.
AI Model2026
Performs zero-shot classification and regression on mixed numerical and categorical tabular data by treating training rows as in-context examples and predicting in a single forward pass. Uses alternating row/column attention and row compression; limited to 10 classes and model weights are non-commercial.
AI Model2026
Thinking-off fine-tune for coding-agent workflows that prioritizes fast next-step decisions, lower token usage and stable multi-turn tool calling. Highlights: MoE 35B base, MTP speculative decoding, SWE-bench 62.4% (300 cases). Best for local agent loops and automated debug cycles; requires disciplined harnessing and schema consistency.
AI Model2026
A large-scale MoE language model for agentic coding and long-context tasks, natively supporting 1M-token context and dynamically activating tens of billions of parameters per token. Uses sparse attention and zero-computation experts to allocate compute per-token; model weights planned for release.
AI Model2026
Provides GGUF/llama.cpp quantized variants of Qwen3.6-27B for local multimodal inference, tuned via online RL to cut average 'thinking' tokens by ≈50% while preserving answer quality; offers Q4_K_M/Q8_0/f16 builds and a separate mmproj for vision input.
AI Infra2026
Streams routed experts from disk and treats storage, RAM and VRAM as a single inference hierarchy so MoE models (hundreds of billions to trillions of parameters) can run on consumer or heterogeneous hardware; single-file C engines with zero runtime deps and optional Vulkan/CUDA/Metal backends, speed governed by placement and disk I/O.
Predicts per-request MoE expert footprints from prefill activations and routes decode requests to workers that maximize expert-locality, lowering decode latency by combining offline K-means partitioning with online locality-band routing and a KV-block–coindexed signature cache.
Xingyu Zheng, Xianglong Liu +5State Key Laboratory of Complex & Critical Software Environment, Beihang University, School of Computer Science and Engineering, Beihang University +6
Accelerates text-to-image diffusion for pretrained flow-matching models using a staged low-to-high-resolution pipeline: fast low-res sampling, pixel-space GAN super-resolution, light latent noising, and short high-res refinement — >10× end-to-end speedups without retraining.
Embodied AI2026
Provides a portable C++ inference runtime to deploy embodied AI models (vision–language–action and world–action) on heterogeneous robot hardware, enabling latency-first batch-1 closed-loop control. Key features include modular multi-rate layers, fused low-latency inference, and extensible head/IO plugins.
Compiles natural-language function specifications into compact, locally-executable neural programs (PAW) that run on a small frozen interpreter; a 4B compiler emits LoRA adapters for a 0.6B runtime to provide offline, low-memory fuzzy text functions.
AI Model2026
Provides a large Mixture-of-Experts instruct LLM (295B total parameters, 21B active, 256K context) optimized for reasoning, long-context retention and agent workflows; open-sourced under Apache-2.0.
AI Video Papers2026
Generates real-time, infinite-length interactive videos of voice-controllable digital characters — 540p at up to 42 FPS on consumer GPUs. Uses TurboDiffusion and TurboServe to maintain temporal coherence without blur or drift, and accepts custom person, anime, or pet images plus selectable voice tones.