mistral-finetune

This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1337

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2.5e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss
2.0914	0.2874	25	1.9959
1.9496	0.5747	50	1.9124
1.8866	0.8621	75	1.8195
1.7543	1.1494	100	1.7156
1.626	1.4368	125	1.6251
1.5435	1.7241	150	1.5451
1.4788	2.0115	175	1.4834
1.3436	2.2989	200	1.4299
1.3102	2.5862	225	1.3759
1.2478	2.8736	250	1.3328
1.1653	3.1609	275	1.2963
1.1127	3.4483	300	1.2587
1.0572	3.7356	325	1.2196
1.0213	4.0230	350	1.1885
0.9496	4.3103	375	1.1625
0.925	4.5977	400	1.1448
0.9109	4.8851	425	1.1337

Framework versions

• PEFT 0.11.2.dev0
• Transformers 4.43.0.dev0
• Pytorch 2.2.1
• Datasets 2.20.0
• Tokenizers 0.19.1