Tag: Transformers

TurboQuant-GPU achieves 5.02x KV cache compression for LLM inference with 0.98 cosine similarity. Learn how this open-source tool outperforms NVIDIA's FP4 formats, works on any GPU, and installs in seconds.

Discover guidance, Microsoft's open-source framework for controlling LLM output with Python code. Learn how constrained generation, token fast-forwarding, and Pydantic-native JSON schemas eliminate parsing failures and cut API costs by 30-70%.

Discover how to train your own 9M-parameter language model from scratch in just 5 minutes using GuppyLM. This complete guide covers architecture, code examples, deployment, and why this tiny fish represents a revolution in AI accessibility.

Discover why spaCy is the industrial-strength NLP library powering production systems across 70+ languages. From blazing Cython speed to transformer integration, learn how to build NLP that actually ships.

Heretic is a fully automatic censorship removal tool for language models that uses advanced directional ablation with TPE optimization. Run one command to decensor AI models while preserving their intelligence. Over 3,000 community models created.

Discover Slowrun, the viral AI benchmark where infinite compute meets 100M tokens. Learn how this Karpathy-endorsed project is revolutionizing language model training by prioritizing learning depth over speed, with real code examples and world record strategies.

Discover pegainfer, a radical pure Rust + CUDA LLM inference engine with zero PyTorch overhead. Achieve 91 tok/s on RTX 5070 Ti, explore custom GPU kernels, and deploy with minimal dependencies. Complete setup guide and benchmarks inside.