Balab2021/Qwen2.5-3B-OASST1-QLoRA_h200-run4-higher-capacity
07
run4highercapacity
QLoRA adapter fine-tuned from Qwen/Qwen2.5-3B-Instruct on OpenAssistant/oasst1.
Final metrics
- train_loss: 1.342932929017136
- eval_loss: 1.497029423713684
- loss_gap: 0.15409649469654818
- eval_perplexity: 4.468395625434644
- train_runtime: 2655.6148
- samples_per_second: 6.821
- global_steps: 1134
