A customizable 30B-parameter Mixture-of-Experts LLM (3B active) in BF16 for low-latency, high-throughput agent workflows; supports speculative decoding (MTP/DSpark/DFlash) and up to 1M-token contexts. Released with open weights and recipes under OpenMDW-1.1, intended for post-training, domain adaptation, and research on NVIDIA GPU stacks.
Multimodal vision-language model optimized for on-device image+text tasks: image captioning, full-page OCR with layout annotation, grounding/bounding-box prediction, and function calling. Built on the LFM2.5-2.6B backbone with a SigLIP2 NaFlex 400M vision encoder and tuned for low-latency, low-memory edge inference.
Provides a machine-readable catalog of 117 AI/AX safety and deployment-readiness diagnostic criteria for assessing model intrinsic and serving/infrastructure risks. Includes MODEL-SCAN and AX-SCAN axes, bilingual source fields, per-item evidence guidance, severity/assurance metadata, and a CC BY-NC 4.0 release-candidate.
Provides a full GGUF quant ladder of an "abliterated" Qwen3.8-27B for local llama.cpp inference — includes every K-quant, embedded MTP speculative head, and optional vision projectors; refusal behavior was reduced at the weight level, so validate before production.
Trains compact conversational agents to adapt at runtime to changing 'Harness' configurations (Skills, Hooks, prompts, tools) using Harness-Aware Training (HAT): Harness-State Augmentation, on-policy distillation, and RL to preserve generality while meeting low-latency deployment constraints.
Local GGUF build of Qwen 3.8 27B with the refusal direction ablated for llama.cpp; includes vision projector (mmproj), MTP speculative head, a 262k context window and multiple quant tiers (Q2–Q8, F16). Research-only release that requires updated llama.cpp and explicit safety layers.
Provides quantized GGUF variants of Qwen3.8-27B with an 'Aggressive' uncensoring profile and an optional HauhauCS FastMTP sidecar to accelerate MTP speculative decoding; includes a BF16 vision projector and K_P quant levels for VRAM/quality trade-offs.
Enables interactive serving of large Mixture-of-Experts (MoE) models on personal machines by adapting offload and execution to measured device bandwidth and agentic workload patterns. Key features include bandwidth-adaptive execution, semantic-aware caching of recurrent state, and an elastic GPU expert cache; supports 20+ MoE models and runs models from ~35B to 753B on consumer/workstation GPUs.
A 9B dense reasoning LLM optimized for single‑GPU deployment and terminal-based coding agents, with long-context support (up to 262,144 tokens) and GGUF/quantized builds for edge/mobile. Strong on coding and agentic benchmarks.
Provides 2-bit quantized weights of Qwen3.8-27B (~10.15 GB) for local deployment, enabling the full 27B parameter model to run on a single 24 GB GPU with long-context support. Delivered as safetensors plus a companion SGLang runtime; measured to match FP8 reference on common benchmarks with small or no quality loss.
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.
A 2B-class causal LLM packaged as a GGUF for local inference; offers 131072-token native long context, XML-style tool-calling support, and is tuned with SFT + RL + OPD using the UltraData family for stronger code, math and agentic abilities.