axolotl-ai-cloud/axolotl

Fine-tuning an open model is way easier than it was a year ago. 5 tools that do the heavy lifting 🧵 • Unsloth: faster, lower-memory fine-tuning, great free notebooks • LLaMA-Factory: fine-tune 100+…

Boost your LLM fine-tuning with Axolotl's latest features. See link below 👇

Long context: 32k-token sequences on a single 80 GB H100 with offloading, and 64k with one more line: activation_offloading: hidden_states

Reallexi AI Model Builder: build models locally, ship them with proof

1 Unsloth Fine-tune LLMs 2x faster with up to 70% less VRAM. Runs on a single consumer GPU

Fundamentals of LLMs, Vectors, Attention, Training karpathy/nanoGPT ( — Minimal, clean implementation for training GPT-style models from scratch. Excellent for understanding attention mechanisms,…

Best GitHub repos for fine-tuning LLMs without melting your GPU: 1. Unsloth 2. Axolotl 3. LLaMA-Factory 4. PEFT 5. TRL 6. Torchtune 7. LitGPT 8. Mergekit 9. SWIFT

Best GitHub repos for fine-tuning LLMs without melting your GPU: 1. Unsloth 2. Axolotl 3. LLaMA-Factory 4. PEFT 5. TRL 6. Torchtune 7. LitGPT 8. Mergekit 9. SWIFT

Best GitHub repos for fine-tuning LLMs without melting your GPU: 1. Unsloth 2. Axolotl 3. LLaMA-Factory 4. PEFT 5. TRL 6. Torchtune 7. LitGPT 8. Mergekit 9. SWIFT

🧵 1/2 If you’ve been hitting the GPU memory wall while trying to fine‑tune LLMs, you’re not alone. The community’s answer? Axolotl – a Python‑first toolkit that squeezes massive models into modest…