The central nervous system and backend orchestrator for Fractal. It handles multi-tenant task scheduling, real-time TFLOPs budgeting, and federated weight aggregation using FedAvg.
-
Updated
Aug 22, 2026 - Python
The central nervous system and backend orchestrator for Fractal. It handles multi-tenant task scheduling, real-time TFLOPs budgeting, and federated weight aggregation using FedAvg.
Open-source AI model aggregation platform with web/admin dashboard, agents, image/video/audio generation, OpenAI-compatible API, wallet billing, and Docker deployment. / 开源 AI 聚合平台,支持大模型工作台、智能体、图片/视频/音频生成、OpenAI 兼容 API、钱包计费、管理后台和 Docker 一键部署。
把 WorkBuddy(多账号/企业版)、Trae CN / TRAE SOLO CN、Qoder CN 三家订阅聚合进同一个 DSH 模型选择器:会话内换模型不丢历史、积分倍率、逐模型思考档位、每日自动签到、读图与 Agent 工具调用。One DSH plugin aggregating WorkBuddy / Trae / Qoder subscriptions into one model picker: mid-session switching, credit multipliers, reasoning effort, daily check-in, vision, tools.
Mixture-of-Agents (MoA) plugin for OpenCode. It fans out a single prompt to multiple models (subagents) in parallel. The primary agent then evaluates and synthesizes the outputs into a unified, high-quality response. Child worker sessions persist and are navigable via the TUI.
Drift-Aware Adaptive Aggregation (DAA) for federated learning on CIFAR-10 under heterogeneous client partitions.
Geometry dynamics and instability in federated LoRA under heterogeneous data distributions (FedGeoX)
A parameter-transfer framework for generalized semiparametric models using model aggregation techniques.
Aggregation Tools for Matrix Population Models in R.
Discover free AI model endpoints across NVIDIA, AMD, Hugging Face, OpenRouter, and models.dev. Generate ready-to-run LiteLLM, NewAPI, and Bifrost configurations. Public read-only API + Agents.md + llms.txt.
Multi-model AI aggregation system with auto-fallback, CCR closed-loop agent, and Self-IDE. Supports DeepSeek, NVIDIA, SiliconFlow, ZenMux, Zhipu, Kimi, MiMo, Mistral.
Data-Enhanced Model Aggregation
Discover free AI model endpoints across NVIDIA, AMD, Hugging Face, OpenRouter, and models.dev. Generate ready-to-run LiteLLM, NewAPI, and Bifrost configurations. Public read-only API + Agents.md + llms.txt.
To associate your repository with the model-aggregation topic, visit your repo's landing page and select "manage topics."