metadata

license: apache-2.0
library_name: peft
tags:
  - generated_from_trainer
base_model: mistralai/Mistral-7B-v0.1
datasets:
  - mbe
metrics:
  - accuracy
model-index:
  - name: Mistral-7B-v0.1_mbe_no
    results: []

Mistral-7B-v0.1_mbe_no

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

Loss: 0.5616
Accuracy: 0.5362

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: 3e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
lr_scheduler_warmup_ratio: 0.03
training_steps: 300

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.5245	0.07	10	0.6507	0.3355
0.6666	0.13	20	0.6464	0.3816
0.6527	0.2	30	0.6427	0.3684
0.6168	0.27	40	0.6321	0.3980
0.6584	0.33	50	0.6182	0.3914
0.586	0.4	60	0.6244	0.4145
0.5924	0.47	70	0.6034	0.4342
0.6069	0.53	80	0.6096	0.4375
0.5999	0.6	90	0.6096	0.4408
0.6206	0.67	100	0.6070	0.4572
0.5793	0.73	110	0.6016	0.4572
0.6208	0.8	120	0.5902	0.4605
0.5622	0.87	130	0.5775	0.4770
0.5502	0.93	140	0.5761	0.4671
0.5958	1.0	150	0.5606	0.4901
0.4558	1.07	160	0.5840	0.4737
0.4411	1.14	170	0.5631	0.4901
0.4144	1.2	180	0.5745	0.5
0.4647	1.27	190	0.5932	0.4605
0.4504	1.34	200	0.5799	0.5099
0.4299	1.4	210	0.6488	0.4934
0.425	1.47	220	0.5704	0.5132
0.4152	1.54	230	0.5582	0.5066
0.425	1.6	240	0.5489	0.5329
0.446	1.67	250	0.5479	0.5197
0.3908	1.74	260	0.5564	0.5164
0.443	1.8	270	0.5419	0.5033
0.4081	1.87	280	0.5948	0.5066
0.3944	1.94	290	0.5547	0.5395
0.4005	2.0	300	0.5616	0.5362

Framework versions

PEFT 0.7.1
Transformers 4.37.2
Pytorch 2.1.2+cu121
Datasets 2.17.1
Tokenizers 0.15.1