Library / SDK
Fine-tuning & Training
XTuner
1Efficiently 2finetune 3large 4models 5and 6MoEs.
the six Ws · specification
W1
Who
Built by InternLM (Shanghai AI Lab).
W2
What
A training engine for QLoRA and full fine-tuning of large and MoE models.
W3
Where
A Python toolkit on GPUs.
W4
When
When fine-tuning on limited GPU memory.
W5
Why
Fine-tunes big models on small hardware.
W6
With
PyTorch and DeepSpeed.
W7
Watch
Apache-2.0, from InternLM; no credentials, runs on your GPUs.
alternatives
works with
InternLMQLoRAMoE