SFTCodePhi-3-mini-4k-instructAPPS3k

This model is a fine-tuned version of microsoft/Phi-3-mini-4k-instruct on the apps dataset. It achieves the following results on the evaluation set:

• Loss: 0.7239

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: 5e-06
• 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: cosine
• lr_scheduler_warmup_steps: 100
• training_steps: 3000

Training results

Training Loss	Epoch	Step	Validation Loss
0.8413	0.6740	200	0.7941
0.7599	1.3479	400	0.7499
0.7385	2.0219	600	0.7340
0.724	2.6959	800	0.7309
0.7263	3.3698	1000	0.7300
0.7194	4.0438	1200	0.7287
0.7186	4.7178	1400	0.7262
0.7204	5.3917	1600	0.7250
0.72	6.0657	1800	0.7243
0.7151	6.7397	2000	0.7244
0.7156	7.4136	2200	0.7238
0.7173	8.0876	2400	0.7240
0.7156	8.7616	2600	0.7240
0.7136	9.4356	2800	0.7239
0.7227	10.1095	3000	0.7239

Framework versions

• PEFT 0.11.0
• Transformers 4.40.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1