Mistral-7B-v0.1_district-court-db

This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0358
• Precision Micro: 0.8142
• Precision Macro: 0.7222
• Recall Micro: 0.8142
• Recall Macro: 0.7126
• F1 Micro: 0.8142
• F1 Macro: 0.7098
• Accuracy: 0.8142

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: 3e-05
• 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: constant
• lr_scheduler_warmup_ratio: 0.03
• training_steps: 1450

Training results

Training Loss	Epoch	Step	Validation Loss	Precision Micro	Precision Macro	Recall Micro	Recall Macro	F1 Micro	F1 Macro	Accuracy
0.1255	0.04	50	0.2459	0.2330	0.0980	0.2330	0.0939	0.2330	0.0773	0.2330
0.1076	0.08	100	0.1451	0.4075	0.1951	0.4075	0.1846	0.4075	0.1681	0.4075
0.066	0.12	150	0.1095	0.5387	0.3493	0.5387	0.2872	0.5387	0.2780	0.5387
0.0699	0.16	200	0.0901	0.6208	0.3837	0.6208	0.3992	0.6208	0.3798	0.6208
0.066	0.2	250	0.0883	0.6104	0.4544	0.6104	0.4312	0.6104	0.4135	0.6104
0.0452	0.24	300	0.0879	0.6877	0.5649	0.6877	0.5135	0.6877	0.5092	0.6877
0.0545	0.28	350	0.0761	0.6764	0.5194	0.6764	0.5288	0.6764	0.5040	0.6764
0.0647	0.32	400	0.0665	0.7340	0.6193	0.7340	0.5252	0.7340	0.5493	0.7340
0.056	0.36	450	0.0514	0.7396	0.6097	0.7396	0.5767	0.7396	0.5672	0.7396
0.0513	0.4	500	0.0479	0.7613	0.6384	0.7613	0.6145	0.7613	0.6020	0.7613
0.0501	0.44	550	0.0502	0.7509	0.6245	0.7509	0.6167	0.7509	0.6075	0.7509
0.0533	0.48	600	0.0481	0.7642	0.6500	0.7642	0.6139	0.7642	0.6073	0.7642
0.0462	0.52	650	0.0473	0.7481	0.5942	0.7481	0.5740	0.7481	0.5679	0.7481
0.0496	0.56	700	0.0419	0.7972	0.6678	0.7972	0.6480	0.7972	0.6518	0.7972
0.0614	0.6	750	0.0489	0.7774	0.6678	0.7774	0.6360	0.7774	0.6308	0.7774
0.0468	0.64	800	0.0443	0.7830	0.6435	0.7830	0.6816	0.7830	0.6494	0.7830
0.0477	0.68	850	0.0420	0.7972	0.7040	0.7972	0.6567	0.7972	0.6663	0.7972
0.0519	0.72	900	0.0463	0.7632	0.6519	0.7632	0.6291	0.7632	0.6292	0.7632
0.0453	0.76	950	0.0429	0.7802	0.6757	0.7802	0.6698	0.7802	0.6564	0.7802
0.0452	0.79	1000	0.0471	0.7377	0.6182	0.7377	0.6300	0.7377	0.6049	0.7377
0.0367	0.83	1050	0.0388	0.7981	0.6857	0.7981	0.6992	0.7981	0.6801	0.7981
0.0377	0.87	1100	0.0382	0.8	0.6636	0.8	0.6698	0.8000	0.6591	0.8
0.0429	0.91	1150	0.0398	0.7953	0.6924	0.7953	0.6441	0.7953	0.6466	0.7953
0.0451	0.95	1200	0.0378	0.7943	0.6713	0.7943	0.6538	0.7943	0.6535	0.7943
0.0347	0.99	1250	0.0413	0.7840	0.6735	0.7840	0.6450	0.7840	0.6331	0.7840
0.0378	1.03	1300	0.0377	0.8047	0.7109	0.8047	0.6387	0.8047	0.6489	0.8047
0.0357	1.07	1350	0.0386	0.8028	0.6899	0.8028	0.6559	0.8028	0.6649	0.8028
0.0418	1.11	1400	0.0368	0.7962	0.7114	0.7962	0.6942	0.7962	0.6910	0.7962
0.0293	1.15	1450	0.0358	0.8142	0.7222	0.8142	0.7126	0.8142	0.7098	0.8142

Framework versions

• PEFT 0.7.1
• Transformers 4.37.2
• Pytorch 2.1.2+cu121
• Datasets 2.17.1
• Tokenizers 0.15.1