Converts an input image into 3D meshes, Gaussian splats, and audio files using Claude skills and external AI APIs.
Archives experiments of AI agents autonomously tuning optimizers, schedules, and hyperparameters to reach target validation loss in fewest steps on a small LM benchmark.
Provides AI agent toolkit with coding CLI, unified LLM API, TUI/web UI libs and Slack bot support.
Automates articulated 3D asset creation by prompting LLMs to generate executable Python code defining parts, geometry and joints.
Streams live traces of coding agent actions including tokens and tool calls to a local UI at localhost:5899.
Implements a minimal AlphaProof-style prover that trains policy-value models and uses MCTS to generate Lean proofs.
Syncs Overleaf projects to local macOS filesystem via Git for offline LaTeX editing.
Pretrains causal language models on 1T tokens of DNA sequences using a hybrid BPE and 6-mer tokenizer for genomic tasks.
Implements a Triton kernel for flash attention that skips fully masked blocks in partially filled binary masks for sparse patterns like causal or sliding-window.
Supplies scripts and datasets to fine-tune Qwen3-8B on an AI-papers wiki using Fireworks Agent.