learning_rate: 5e-05
train_batch_size: 1
eval_batch_size: 1
seed: 12345
distributed_type: multi-GPU
num_devices: 8
total_train_batch_size: 8
total_eval_batch_size: 8
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Roc Auc	Best Threshold	Tpr At Best Threshold	Fpr At Best Threshold	Accuracy At Best Threshold	Avg Score True Positives	Avg Score True Negatives
No log	0	0	40.8364	0.6060	0.0100	0.4225	0.1364	0.5269	0.1403	0.0493
29.1305	0.0658	5	39.9649	0.6018	0.0111	0.4225	0.1591	0.5215	0.1440	0.0544
30.9881	0.1316	10	37.7798	0.6002	0.0063	0.4930	0.25	0.5538	0.1645	0.0601
28.378	0.1974	15	31.6639	0.6112	0.0405	0.4648	0.2045	0.5430	0.2268	0.0812
26.1955	0.2632	20	19.1034	0.6216	0.7677	0.3732	0.0909	0.5	0.4480	0.2411
25.1083	0.3289	25	13.8341	0.6338	0.9317	0.4437	0.0909	0.5538	0.5938	0.4443
16.2399	0.3947	30	11.1298	0.6380	0.9910	0.4648	0.1136	0.5645	0.7250	0.6482
20.1369	0.4605	35	10.5893	0.6610	0.9991	0.4225	0.0682	0.5430	0.8160	0.7614
5.2696	0.5263	40	10.4212	0.6660	0.9994	0.4155	0.0682	0.5376	0.8449	0.7797
16.6356	0.5921	45	9.9733	0.6757	0.9986	0.5423	0.1591	0.6129	0.8745	0.8105
12.458	0.6579	50	9.7818	0.6728	0.9993	0.4789	0.1136	0.5753	0.8951	0.8368
35.8137	0.7237	55	9.4100	0.6887	0.9993	0.4930	0.1136	0.5860	0.9182	0.8492
4.1218	0.7895	60	7.8306	0.6962	0.9846	0.6549	0.2955	0.6667	0.8823	0.7783
3.3694	0.8553	65	7.0107	0.7043	0.8485	0.6620	0.2955	0.6720	0.7550	0.5274
19.8986	0.9211	70	8.0077	0.7094	0.9012	0.5	0.1136	0.5914	0.6576	0.3670
3.5818	0.9868	75	7.5069	0.7223	0.8454	0.5563	0.1364	0.6290	0.6770	0.3726
5.2156	1.0526	80	6.7758	0.7316	0.9070	0.5775	0.1591	0.6398	0.7328	0.4362
4.6982	1.1184	85	6.1916	0.7362	0.9689	0.5704	0.1364	0.6398	0.8005	0.5451
7.6836	1.1842	90	6.2596	0.7485	0.9949	0.5211	0.0909	0.6129	0.8797	0.6767
6.414	1.25	95	6.4384	0.7655	0.9840	0.6620	0.2045	0.6935	0.9082	0.7268
4.827	1.3158	100	5.9971	0.7670	0.9933	0.5704	0.1136	0.6452	0.8901	0.6661
4.8294	1.3816	105	5.4045	0.7782	0.8228	0.7465	0.2727	0.7419	0.8272	0.5140
6.0952	1.4474	110	5.7227	0.7742	0.8800	0.6197	0.1818	0.6667	0.7607	0.4081
4.7037	1.5132	115	5.8822	0.7866	0.8539	0.6056	0.1364	0.6667	0.7290	0.3426
4.3716	1.5789	120	5.4608	0.7926	0.9258	0.6127	0.1364	0.6720	0.7776	0.4070
0.5355	1.6447	125	5.2155	0.7898	0.8522	0.7465	0.25	0.7473	0.8383	0.5165
10.7298	1.7105	130	5.1987	0.7931	0.9816	0.5704	0.0909	0.6505	0.8458	0.5196
0.1842	1.7763	135	5.2841	0.7952	0.8697	0.7394	0.25	0.7419	0.8304	0.4836
18.9794	1.8421	140	5.2954	0.7964	0.8339	0.7183	0.2273	0.7312	0.8191	0.4570
4.7871	1.9079	145	5.5306	0.7980	0.6549	0.7606	0.25	0.7581	0.7749	0.3739
4.9771	1.9737	150	6.3241	0.7892	0.4494	0.7676	0.2727	0.7581	0.7173	0.3077
0.2814	2.0395	155	6.3730	0.7887	0.5641	0.7113	0.2273	0.7258	0.7133	0.3087
4.8055	2.1053	160	6.0185	0.7963	0.3208	0.8028	0.2955	0.7796	0.7313	0.3229
2.877	2.1711	165	5.9060	0.7946	0.4766	0.7676	0.2727	0.7581	0.7426	0.3349
6.9611	2.2368	170	6.0269	0.7923	0.4455	0.7676	0.2727	0.7581	0.7374	0.3284
8.9122	2.3026	175	5.6056	0.8003	0.4961	0.7887	0.2955	0.7688	0.7687	0.3630
5.4598	2.3684	180	5.3411	0.7954	0.6077	0.7676	0.2727	0.7581	0.7885	0.3923
1.8256	2.4342	185	5.2602	0.8055	0.4321	0.8451	0.3409	0.8011	0.7986	0.3939
5.2833	2.5	190	5.3086	0.7934	0.7931	0.7324	0.2273	0.7419	0.8075	0.4348
1.4256	2.5658	195	5.2967	0.7971	0.7090	0.7676	0.2727	0.7581	0.8122	0.4368
3.4508	2.6316	200	5.2114	0.7999	0.7122	0.7746	0.2955	0.7581	0.8221	0.4495
9.6506	2.6974	205	5.1599	0.7999	0.7725	0.7676	0.2727	0.7581	0.8225	0.4471
0.3497	2.7632	210	5.2210	0.7990	0.5467	0.8310	0.3182	0.7957	0.8185	0.4388
0.921	2.8289	215	5.2554	0.7972	0.5775	0.8310	0.3409	0.7903	0.8222	0.4566
0.373	2.8947	220	5.1279	0.8042	0.5946	0.8239	0.3182	0.7903	0.8246	0.4374
0.1713	2.9605	225	5.2345	0.7991	0.8622	0.7113	0.2273	0.7258	0.8217	0.4480

Framework versions

PEFT 0.18.1
Transformers 4.57.6
Pytorch 2.9.1+cu128
Datasets 4.5.0
Tokenizers 0.22.2

leobianco/bosch_RM_Qwen_S12345_LLM_false_STRUCT_false_epo3_lr5e-5_r8_2602040939

bosch_RM_Qwen_S12345_LLM_false_STRUCT_false_epo3_lr5e-5_r8_2602040939

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

leobianco/
bosch_RM_Qwen_S12345_LLM_false_STRUCT_false_epo3_lr5e-5_r8_2602040939