FlashML-org/FreeToken
Runs open-weight mixture-of-experts models locally on consumer hardware.
View on index · View in 3D Map
Related repos
- MLA decoding kernel for Hopper GPUs
- open-weight GPT models for reasoning and agent tasks
- Open-source platform for evaluating and optimizing LLM apps
- Hosts open-source models as APIs
- Benchmark for LLMs on software engineering tasks
- Toolkit for research on retrieval augmented generation
- LLM training framework for Apple Silicon
- Runs hundreds of large models on a single GPU by loading them from SSD to VRAM up to 10x faster than alternative loaders, achieving coldstarts under 2 seconds.
- Agent for research across model providers and MCP servers
- library for training large models across any modality, config-driven and modular, from Apple Research
