lmind_nq_train6000_eval6489_v1_doc_qa_v3_5e-5_lora2

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the tyzhu/lmind_nq_train6000_eval6489_v1_doc_qa_v3 dataset. It achieves the following results on the evaluation set:

• Loss: 6.7918
• Accuracy: 0.1930

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: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• total_eval_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 50.0

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
1.3891	1.0	529	0.6138	1.3015
1.3633	2.0	1058	0.6166	1.2855
1.2929	3.0	1587	0.6177	1.2954
1.2361	4.0	2116	0.6045	1.3489
1.1856	5.0	2645	0.6125	1.3968
1.1098	6.0	3174	0.6115	1.4721
1.0753	7.0	3703	0.6076	1.5798
1.0048	8.0	4232	0.6084	1.6042
0.9456	9.0	4761	0.5977	1.6843
0.8766	10.0	5290	0.6051	1.7829
0.8273	11.0	5819	0.6043	1.8060
0.7755	12.0	6348	0.6019	1.8729
0.715	13.0	6877	0.6017	1.9620
0.6804	14.0	7406	0.6009	2.0030
0.6277	15.0	7935	0.5998	2.0528
0.5733	16.0	8464	0.6012	2.0475
0.5409	17.0	8993	0.5749	2.0920
0.5024	18.0	9522	0.5986	2.1207
0.4699	19.0	10051	0.5993	2.1108
0.4367	20.0	10580	0.6005	2.1089
0.857	21.0	11109	0.5983	2.0215
3.7434	22.0	11638	0.2233	10.1186
7.7259	23.0	12167	0.1986	7.5379
4.2204	24.0	12696	0.5345	2.1568
0.7385	25.0	13225	0.5963	1.8229
1.1473	26.0	13754	0.5788	1.7570
2.0182	27.0	14283	0.5573	1.7293
2.2707	28.0	14812	0.4956	2.7017
4.1792	29.0	15341	0.3070	5.8288
7.7703	30.0	15870	0.1922	7.6619
7.7034	31.0	16399	0.1913	7.7003
7.9533	32.0	16928	0.1899	7.8667
7.8634	33.0	17457	0.1897	7.8134
7.8584	34.0	17986	0.1882	7.6760
7.824	35.0	18515	0.1888	7.7083
7.7446	36.0	19044	0.1888	7.6626
7.6708	37.0	19573	0.1886	7.5529
7.6733	38.0	20102	0.1903	7.5704
7.6271	39.0	20631	0.1949	7.5363
7.5886	40.0	21160	0.2130	7.4684
7.5514	41.0	21689	0.2077	7.4223
7.5205	42.0	22218	0.1946	7.3508
7.4577	43.0	22747	0.1951	7.1785
7.5021	44.0	23276	0.2092	6.6226
7.1133	45.0	23805	0.1994	6.4100
6.9682	46.0	24334	0.2250	6.3553
6.8891	47.0	24863	0.2224	6.3128
6.8621	48.0	25392	0.23	6.2465
6.8176	49.0	25921	0.2561	6.1966
6.9473	50.0	26450	0.1930	6.7918

Framework versions

• Transformers 4.34.0
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.14.1