metadata

license: other
library_name: peft
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - accuracy
base_model: microsoft/phi-2
model-index:
  - name: Phi2-Seq-classification-LoRa
    results: []

Phi2-Seq-classification-LoRa

This model is a fine-tuned version of microsoft/phi-2 on the None dataset. It achieves the following results on the evaluation set:

Loss: 2.5015
Precision: 0.9783
Recall: 0.8182
F1-score: 0.8911
Accuracy: 0.9236

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: 0.0002
train_batch_size: 1
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
lr_scheduler_warmup_ratio: 0.03
training_steps: 3000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1-score	Accuracy
0.3914	0.7	100	0.5650	0.8571	0.6545	0.7423	0.8264
0.0065	1.39	200	0.8945	0.9375	0.8182	0.8738	0.9097
0.2111	2.09	300	0.6418	0.9388	0.8364	0.8846	0.9167
0.1823	2.78	400	0.5655	0.9773	0.7818	0.8687	0.9097
0.145	3.48	500	0.9232	0.9545	0.7636	0.8485	0.8958
0.1095	4.17	600	0.8011	0.9583	0.8364	0.8932	0.9236
0.0	4.87	700	0.9774	0.9796	0.8727	0.9231	0.9444
0.4204	5.57	800	0.6787	0.96	0.8727	0.9143	0.9375
0.0068	6.26	900	1.1916	0.9783	0.8182	0.8911	0.9236
0.0004	6.96	1000	0.9468	0.9783	0.8182	0.8911	0.9236
0.0	7.65	1100	1.2818	0.9787	0.8364	0.9020	0.9306
0.0	8.35	1200	1.2332	0.9787	0.8364	0.9020	0.9306
0.0	9.04	1300	1.3078	0.9583	0.8364	0.8932	0.9236
0.0	9.74	1400	1.9713	0.9773	0.7818	0.8687	0.9097
0.0001	10.43	1500	1.5254	0.9773	0.7818	0.8687	0.9097
0.0	11.13	1600	1.8045	0.9592	0.8545	0.9038	0.9306
0.0011	11.83	1700	2.7594	0.9574	0.8182	0.8824	0.9167
0.0	12.52	1800	1.9378	0.9592	0.8545	0.9038	0.9306
0.0	13.22	1900	1.8710	0.9592	0.8545	0.9038	0.9306
0.0	13.91	2000	1.8685	0.9592	0.8545	0.9038	0.9306
0.0	14.61	2100	1.8683	0.9592	0.8545	0.9038	0.9306
0.0384	15.3	2200	2.3028	0.9592	0.8545	0.9038	0.9306
0.001	16.0	2300	1.7787	0.9216	0.8545	0.8868	0.9167
0.0	16.7	2400	1.8615	0.9787	0.8364	0.9020	0.9306
0.8746	17.39	2500	2.6710	0.9756	0.7273	0.8333	0.8889
0.0	18.09	2600	1.0475	0.9412	0.8727	0.9057	0.9306
0.0	18.78	2700	2.3325	0.9778	0.8	0.8800	0.9167
0.0001	19.48	2800	1.3658	0.94	0.8545	0.8952	0.9236
0.0	20.17	2900	2.4668	0.9783	0.8182	0.8911	0.9236
0.0	20.87	3000	2.5015	0.9783	0.8182	0.8911	0.9236

Framework versions

PEFT 0.7.1
Transformers 4.36.2
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.15.0