Undi95/Mistral-7B-roleplay_alpaca-lora

roleplay_alpaca

This model was trained from scratch on the TokenBender/roleplay_alpaca dataset. It achieves the following results on the evaluation set:

• Loss: 1.2973

Model description

Dataset trained on mistralai/Mistral-7B-v0.1

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.00065
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 10
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
1.5404	0.03	20	1.2764
1.4523	0.06	40	1.2571
1.5608	0.09	60	1.2465
1.5921	0.12	80	1.2575
1.6043	0.15	100	1.2432
1.5496	0.18	120	1.2504
1.348	0.21	140	1.2452
1.4638	0.24	160	1.2661
1.5733	0.27	180	1.2548
1.5397	0.3	200	1.2674
1.6154	0.33	220	1.2626
1.5058	0.36	240	1.2672
1.3974	0.4	260	1.2659
1.6654	0.43	280	1.2648
1.8051	0.46	300	1.2585
1.7487	0.49	320	1.2736
1.3612	0.52	340	1.2717
1.5048	0.55	360	1.2809
1.7134	0.58	380	1.2885
1.5524	0.61	400	1.2805
1.3705	0.64	420	1.2860
1.4335	0.67	440	1.2896
1.3642	0.7	460	1.2911
1.6546	0.73	480	1.2888
1.5345	0.76	500	1.2973
1.5968	0.79	520	1.2885
1.5694	0.82	540	1.2939
1.5474	0.85	560	1.2892
1.6981	0.88	580	1.2949
1.5451	0.91	600	1.2886
1.5845	0.94	620	1.2941
1.5143	0.97	640	1.2973

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.14.0

If you want to support me, you can here.