Commits

Commits on Mar 15, 2026

Merge pull request #106 from tensorlink-lab/testnet-v2
mattjhawken
authored

Commits on Mar 8, 2026

Increased connection.py parameters for faster P2P comms.
mattjhawken
committed

Commits on Mar 7, 2026

Switched custom serialization to safetensors save. Fixed key error in /models api endpoint.
mattjhawken
committed

Commits on Mar 4, 2026

Added hidden buffer loading for worker and host modules.
mattjhawken
committed

Commits on Mar 3, 2026

Commits on Mar 2, 2026

Commits on Feb 26, 2026

Commits on Feb 24, 2026

Improved layer module loading fallbacks for mismatched keys.
mattjhawken
committed

Commits on Feb 23, 2026

Commits on Feb 21, 2026

Fixed memory estimation bug in worker. Modularized model skeleton loading for use in graphing, worker, and module. Fixed llama model path resolution.
mattjhawken
committed

Commits on Feb 20, 2026

Made enable_grad conditional on training for worker and validator. Added DynamicCache handling for custom attach/detach functions.
mattjhawken
committed

Commits on Feb 19, 2026

Removed auto-loading model feature (temporarily).
mattjhawken
committed

Commits on Feb 10, 2026

Commits on Feb 9, 2026

Commits on Jan 30, 2026

Ensured layer group modules are loaded to device on worker.
mattjhawken
committed

Commits on Jan 29, 2026