jerseyjerry/task-4-microsoft-Phi-3-mini-4k-instruct-20250222

lora

This model is a fine-tuned version of microsoft/Phi-3-mini-4k-instruct on the flock_task4_tranning dataset. It achieves the following results on the evaluation set:

• Loss: 1.1333

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-05
• train_batch_size: 2
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• total_eval_batch_size: 2
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
1.1	0.0501	50	1.2970
1.0774	0.1002	100	1.2504
0.9563	0.1502	150	1.2521
1.0242	0.2003	200	1.2253
1.0038	0.2504	250	1.2113
0.9858	0.3005	300	1.1920
0.8694	0.3505	350	1.1918
0.9174	0.4006	400	1.1884
0.9653	0.4507	450	1.1870
0.9136	0.5008	500	1.1768
0.9014	0.5508	550	1.1673
0.9203	0.6009	600	1.1558
0.8902	0.6510	650	1.1679
0.9018	0.7011	700	1.1489
0.937	0.7511	750	1.1577
0.8984	0.8012	800	1.1463
0.8607	0.8513	850	1.1517
0.8698	0.9014	900	1.1436
0.9661	0.9514	950	1.1479
0.672	1.0010	1000	1.1459
0.8162	1.0511	1050	1.1374
0.8477	1.1012	1100	1.1434
0.9039	1.1512	1150	1.1394
0.8361	1.2013	1200	1.1434
0.8091	1.2514	1250	1.1391
0.7854	1.3015	1300	1.1392
0.7716	1.3515	1350	1.1403
0.8637	1.4016	1400	1.1337
0.8491	1.4517	1450	1.1392
0.9037	1.5018	1500	1.1325
0.8698	1.5518	1550	1.1371
0.7614	1.6019	1600	1.1332
0.7492	1.6520	1650	1.1350
0.8217	1.7021	1700	1.1321
0.8261	1.7521	1750	1.1323
0.8286	1.8022	1800	1.1325
0.8208	1.8523	1850	1.1335
0.7937	1.9024	1900	1.1338
0.837	1.9524	1950	1.1342

Framework versions

• PEFT 0.12.0
• Transformers 4.48.3
• Pytorch 2.6.0+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0