titans-pytorch
Unofficial implementation of Titans, SOTA memory for transformers, in Pytorch
- 2k
- Python
- MIT
- 3 wk ago
Unofficial implementation of Titans, SOTA memory for transformers, in Pytorch
Collection of awesome test-time (domain/batch/instance) adaptation methods
Implementation of RoboTTT proposed by Yunfan Jiang et al. of Stanford and Nvidia
A tiny test-time-training state-space model with a transactional safety harness.
Full-context associative memory via compact equilibrium solves, with linear sequence scaling and PyTorch/Triton acceleration.
Minimal PyTorch implementation of Titans' test-time neural memory and its MAC/MAG/MAL architectures (arXiv:2501.00663)
Run RL-driven autoresearch on Apple Silicon — local-first MLX version of TTT-Discover AutoResearch
Technical report on the Agnes Sys1 1.0 decision model, architecture, test-time training methodology, and reported capability evaluation.
TTTLab — Interactive Test-Time Training & Adaptive Inference Laboratory. 6 modules: TTT Inner Loop Engine, Sequence Memory Dynamics, Context Length Scaling, Self-Supervised Inner World, Distribution Shift Adaptation, Architecture Arena. Zero dependencies.
Step-by-step implementation of next-generation LLM components. Deep dive into Quantization, TTT layers, Differentiable Logic, FlashAttention tiling, and context extension, culminating in an end-to-end hardware-friendly Mini-LLM.