dalyaff/phi2-QA-Arabic-phi-darebah-2

phi-2

This model is a fine-tuned version of microsoftl on the dalyaff/darebah dataset. It achieves the following results on the evaluation set:

• Loss: 0.7781

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: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5
• training_steps: 2500

Training results

Training Loss	Epoch	Step	Validation Loss
1.1114	0.89	100	1.0073
0.8772	1.78	200	0.8791
0.7638	2.67	300	0.8327
0.7518	3.56	400	0.8077
0.6624	4.44	500	0.7896
0.6386	5.33	600	0.7826
0.6161	6.22	700	0.7677
0.6053	7.11	800	0.7669
0.5725	8.0	900	0.7640
0.5569	8.89	1000	0.7705
0.5303	9.78	1100	0.7691
0.546	10.67	1200	0.7671
0.5331	11.56	1300	0.7696
0.5142	12.44	1400	0.7670
0.5037	13.33	1500	0.7713
0.4938	14.22	1600	0.7686
0.4879	15.11	1700	0.7733
0.4743	16.0	1800	0.7730
0.4705	16.89	1900	0.7739
0.4942	17.78	2000	0.7770
0.4669	18.67	2100	0.7709
0.462	19.56	2200	0.7788
0.4667	20.44	2300	0.7783
0.4638	21.33	2400	0.7754
0.4512	22.22	2500	0.7781

Framework versions

• PEFT 0.7.2.dev0
• Transformers 4.38.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0