FareedKhan-dev/train-llm-from-scratch
Trains billion-parameter LLMs from scratch on a single GPU
View on index · View in 3D Map
Related repos
- Single-GPU training for 100B+ parameter LLMs
- Matches LLM models to hardware specifications
- Trains models exceeding GPU memory
- Code for building a GPT-like LLM from scratch in PyTorch
- Harness for training LLMs on a single GPU node
- Builds GPT-like LLMs from scratch in PyTorch
- Course on building LLM applications from scratch
- Automates GPU kernel generation with LLMs and agents
- Hardware-accelerated GPU terminal emulator
- Open-source tools for running and improving LLM applications—gateway, observability, data feedback, optimization
