Tag
Explore by tags
A vision-language-action foundation model and reference stack for generalized humanoid and cross-embodiment robot manipulation. Provides pretrained checkpoints, demo datasets, and tooling for fine-tuning, evaluation, and deployment (ONNX/TensorRT); released as Early Access.
Trains multi-step LLM agents with reinforcement learning (GRPO) on your own tasks, wrapping existing agent code behind an OpenAI-compatible client. Its RULER mode scores trajectories with an LLM judge, so there's no reward function to hand-write.
Provides 7×24 automated customer service for the Xianyu marketplace with multi-expert routing, context-aware dialogue, and a laddered bargaining system. Built in Python and designed to run against an LLM provider with browser-cookie integration for web interactions.
Transforms unstructured financial content—papers, news, blogs, and filings—into a queryable semantic knowledge graph for retrieval-augmented research. Combines domain-tuned LLMs, embedding-based search, and modular ingestion pipelines; aimed at quant research teams and institutional workflows.
Connects AI coding agents (Cursor, Claude Code) to Figma through a WebSocket bridge, letting an agent read a design and edit it programmatically. Includes a Figma plugin and 40+ MCP tools for text, styling, components, and bulk edits.
Splits LLM inference into separate prefill and decode GPU pools, then routes requests with KV-cache awareness to cut redundant recomputation. Reports up to 30x throughput on DeepSeek-R1 (GB200 NVL72) and works across TensorRT-LLM, vLLM, and SGLang.
Real-time DETR detector on a DINOv2 backbone, covering detection, segmentation, and keypoints. Ships in six sizes (Nano to 2XL), beats YOLO on the COCO speed-accuracy curve, and transfers better to non-COCO real-world domains.
Connects Claude (via the Model Context Protocol) to Ableton Live so the LLM can create and edit tracks, clips, instruments, and control playback through a socket-based MCP server and an Ableton MIDI Remote Script.
Gives an LLM a browser via Playwright's accessibility tree instead of screenshots, so the model reads structured snapshots, not pixels. Actions target named elements deterministically, cutting token use and removing any need for a vision model.
Exposes AWS services to AI agents over the Model Context Protocol — querying databases, provisioning infrastructure with CDK/EKS/Lambda, and pulling live AWS docs. Each domain ships as its own server, so agents wire in only what a task needs.
Gives coding agents symbol-level codebase access via language servers (LSP), turning cross-file renames, reference lookups, and edits into precise operations instead of fragile text search. Runs as an MCP server spanning 40+ languages.