code-llama-instruct-7b-text-to-sparql-axiom

This model is a fine-tuned version of codellama/CodeLlama-7b-Instruct-hf on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1333

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 800
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.3919	0.0710	20	1.3906
0.8712	0.1421	40	0.4591
0.2672	0.2131	60	0.2378
0.213	0.2842	80	0.2065
0.1697	0.3552	100	0.2208
0.2068	0.4263	120	0.1886
0.1808	0.4973	140	0.1843
0.2073	0.5684	160	0.1812
0.1833	0.6394	180	0.1735
0.1556	0.7105	200	0.1836
0.1813	0.7815	220	0.1688
0.166	0.8526	240	0.1642
0.1773	0.9236	260	0.1609
0.1514	0.9947	280	0.1597
0.1592	1.0657	300	0.1581
0.1632	1.1368	320	0.1552
0.1601	1.2078	340	0.1554
0.1529	1.2789	360	0.1523
0.1352	1.3499	380	0.1528
0.1601	1.4210	400	0.1496
0.1523	1.4920	420	0.1482
0.1568	1.5631	440	0.1482
0.1598	1.6341	460	0.1461
0.1432	1.7052	480	0.1471
0.158	1.7762	500	0.1430
0.1479	1.8472	520	0.1422
0.1488	1.9183	540	0.1429
0.1422	1.9893	560	0.1397
0.149	2.0604	580	0.1391
0.1352	2.1314	600	0.1381
0.1357	2.2025	620	0.1389
0.1519	2.2735	640	0.1369
0.1321	2.3446	660	0.1367
0.1381	2.4156	680	0.1361
0.1362	2.4867	700	0.1349
0.1329	2.5577	720	0.1351
0.1457	2.6288	740	0.1340
0.1267	2.6998	760	0.1336
0.1433	2.7709	780	0.1335
0.1343	2.8419	800	0.1333

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.10.1
• Tokenizers 0.19.1