Llama3_ALL_BCE_translations_19_shuffled_special_tokens

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.4776
• F1 Macro 0.1: 0.0818
• F1 Macro 0.15: 0.0922
• F1 Macro 0.2: 0.1027
• F1 Macro 0.25: 0.1130
• F1 Macro 0.3: 0.1230
• F1 Macro 0.35: 0.1336
• F1 Macro 0.4: 0.1440
• F1 Macro 0.45: 0.1551
• F1 Macro 0.5: 0.1663
• F1 Macro 0.55: 0.1778
• F1 Macro 0.6: 0.1879
• F1 Macro 0.65: 0.1987
• F1 Macro 0.7: 0.2090
• F1 Macro 0.75: 0.2178
• F1 Macro 0.8: 0.2211
• F1 Macro 0.85: 0.2205
• F1 Macro 0.9: 0.2010
• F1 Macro 0.95: 0.1457
• Threshold 0: 0.65
• Threshold 1: 0.75
• Threshold 2: 0.7
• Threshold 3: 0.85
• Threshold 4: 0.8
• Threshold 5: 0.85
• Threshold 6: 0.8
• Threshold 7: 0.8
• Threshold 8: 0.85
• Threshold 9: 0.75
• Threshold 10: 0.85
• Threshold 11: 0.8
• Threshold 12: 0.85
• Threshold 13: 0.95
• Threshold 14: 0.85
• Threshold 15: 0.75
• Threshold 16: 0.85
• Threshold 17: 0.8
• Threshold 18: 0.9
• 0: 0.0619
• 1: 0.1388
• 2: 0.1978
• 3: 0.1328
• 4: 0.2961
• 5: 0.3489
• 6: 0.3179
• 7: 0.1268
• 8: 0.2043
• 9: 0.3668
• 10: 0.3216
• 11: 0.3669
• 12: 0.1276
• 13: 0.1205
• 14: 0.2264
• 15: 0.1576
• 16: 0.3078
• 17: 0.3722
• 18: 0.125
• Max F1: 0.2211
• Mean F1: 0.2273

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: 8
• eval_batch_size: 8
• seed: 2024
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Macro 0.1	F1 Macro 0.15	F1 Macro 0.2	F1 Macro 0.25	F1 Macro 0.3	F1 Macro 0.35	F1 Macro 0.4	F1 Macro 0.45	F1 Macro 0.5	F1 Macro 0.55	F1 Macro 0.6	F1 Macro 0.65	F1 Macro 0.7	F1 Macro 0.75	F1 Macro 0.8	F1 Macro 0.85	F1 Macro 0.9	F1 Macro 0.95	Threshold 0	Threshold 1	Threshold 2	Threshold 3	Threshold 4	Threshold 5	Threshold 6	Threshold 7	Threshold 8	Threshold 9	Threshold 10	Threshold 11	Threshold 12	Threshold 13	Threshold 14	Threshold 15	Threshold 16	Threshold 17	Threshold 18	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	Max F1	Mean F1
3.3824	1.0	5595	4.3847	0.0700	0.0761	0.0818	0.0877	0.0936	0.1000	0.1064	0.1134	0.1196	0.1265	0.1327	0.1381	0.1432	0.1483	0.1465	0.1417	0.1291	0.0836	0.65	0.9	0.85	0.9	0.75	0.6	0.8	0.75	0.9	0.9	0.9	0.85	0.9	0.0	0.85	0.75	0.6	0.6	0.9	0.0649	0.0879	0.1603	0.0899	0.2589	0.2876	0.2683	0.1036	0.1245	0.2856	0.2387	0.3033	0.0726	0.0	0.1779	0.1109	0.2192	0.2743	0.0641	0.1483	0.1680
2.4859	2.0	11190	1.7537	0.0881	0.0994	0.1111	0.1210	0.1310	0.1401	0.1472	0.1541	0.1607	0.1676	0.1697	0.1731	0.1768	0.1761	0.1713	0.1575	0.1365	0.0927	0.55	0.7	0.85	0.8	0.4	0.35	0.95	0.75	0.7	0.85	0.8	0.65	0.8	0.95	0.8	0.7	0.85	0.6	0.75	0.0534	0.1241	0.1924	0.1020	0.2738	0.3163	0.3072	0.1109	0.1793	0.3414	0.2889	0.3332	0.0831	0.0870	0.2137	0.1305	0.2881	0.3396	0.1254	0.1768	0.2048
1.7561	3.0	16785	1.4633	0.0840	0.0954	0.1062	0.1164	0.1271	0.1382	0.1485	0.1597	0.1713	0.1809	0.1895	0.1976	0.2056	0.2113	0.2115	0.1995	0.1805	0.1184	0.6	0.75	0.75	0.95	0.8	0.7	0.9	0.8	0.8	0.7	0.8	0.8	0.9	0.95	0.75	0.8	0.7	0.7	0.8	0.0581	0.1395	0.1946	0.1235	0.2818	0.3391	0.3151	0.1202	0.1997	0.3656	0.3056	0.3630	0.1340	0.1087	0.2272	0.1482	0.2953	0.3589	0.1233	0.2115	0.2211
1.2709	4.0	22380	1.4776	0.0818	0.0922	0.1027	0.1130	0.1230	0.1336	0.1440	0.1551	0.1663	0.1778	0.1879	0.1987	0.2090	0.2178	0.2211	0.2205	0.2010	0.1457	0.65	0.75	0.7	0.85	0.8	0.85	0.8	0.8	0.85	0.75	0.85	0.8	0.85	0.95	0.85	0.75	0.85	0.8	0.9	0.0619	0.1388	0.1978	0.1328	0.2961	0.3489	0.3179	0.1268	0.2043	0.3668	0.3216	0.3669	0.1276	0.1205	0.2264	0.1576	0.3078	0.3722	0.125	0.2211	0.2273

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.2.2+cu121
• Datasets 2.18.0
• Tokenizers 0.19.1