AI Dataset2026
A synthetic, verifiable-first agentic training corpus with 19,072 training traces and 2,135 held-out evaluation rows. Provides per-turn visible reasoning, real sandboxed tool executions, 13 verifiable task families, and NeMo Gym / RL-ready reward contracts for SFT and RL workflows.
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.
AI Client2026
Exposes models available in the WorkBuddy / WorkBuddy AI desktop apps to DeepSeek Harness with zero configuration. Reuses local app login and balances, shows separate domestic/international model groups, supports image inputs and inference-tier selection; intended for personal research use.
AI Client2026
Separates host-layer policy from model behavior in DeepSeek Harness by installing reversible local patches and editable prompt-injects for red-team evaluation of LLM hosts. Includes a Clean/Drill side dock, rule sets, skill import, and Apply/Restore controls for official web and desktop hosts.
AI Agent Papers2026
Synthesizes, repairs, and self-evolves task-adaptive agent harnesses on demand for off-the-shelf LLM agents, using a trainable harness-intelligence model that distills signals from past configurations. Demonstrates consistent performance gains across benchmarks and model families by producing four-module, composable harnesses.
AI Model2026
An experimental multimodal model that adds visual understanding to DeepSeek-V4-Flash: accepts text+image inputs and returns text analyses. Improves vision-dependent agent workflows while maintaining comparable text-only performance; released under an MIT license on Hugging Face.
AI Agent Papers2026
Turns each research paper into a training environment to generate verifiable research plans by synthesizing questions from goals/background and deriving evaluation criteria from methods/experiments. Key features: four-stage extraction that reduces criterion leakage to 3.7%, a two-stage rubric-centered training (self-distillation then GRPO), and the PaperGym-20k corpus with two held-out benchmarks.
AI Model2026
A multimodal Mixture-of-Experts foundation model with a million-token context window; uses a causal encoder–decoder layout and aggressive KV-cache compression (~890 bytes/token) to limit per-token activation to 8B/16B—designed for long-context, agentic, and multimodal workloads.
AI Model2026
Provides a drop-in checkpoint of DeepSeek-V4.1-Flash with weight-level abliteration that removes safety guardrails to produce uncensored outputs; preserves vision, MoE routing, 1M-token context and native FP8 quantization. Intended for advanced self-hosted deployment; requires large NVLink GPU domains and careful serving setup.
Describes DeepSeek-V4.1-Flash, a 552B-parameter multimodal Mixture-of-Experts LLM that supports up to 1M-token contexts and reduces KV cache footprint using Compressed Sparse Attention 2, FP4 KV caching, and SWA Bounded Replay to lower HBM/SSD and prefill costs.
Hangyu Ran, Yuhao Zheng +3Ant Group, The Hong Kong University of Science and Technology (Guangzhou) +2
Converts natural-language SVG instructions into instance-aware six-item rubrics and uses a vision–language judge to score rendered SVGs item-by-item as dense RL rewards. Produces stronger human alignment and lifts benchmark rubric scores substantially on MMSVG datasets.
AI Agent Papers2026
Runs an open-source personal software agent across a user's devices to perform actions, keep readable provenance-backed memories, and coordinate via a self-hostable relay. Ships minimal end-to-end clients (mobile, desktop, web) under GPL-3.0; model-agnostic and designed for inspectability and device