lmind_nq_train6000_eval6489_v1_reciteonly_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_reciteonly_qa_v3 dataset. It achieves the following results on the evaluation set:

• Loss: 2.9168
• Accuracy: 0.6388

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
1.3988	1.0	187	0.6661	1.2107
1.1977	2.0	375	0.6676	1.1987
1.1458	3.0	562	0.6676	1.1964
1.0707	4.0	750	0.6666	1.2088
1.0066	5.0	937	0.6652	1.2342
0.9353	6.0	1125	0.6628	1.2646
0.8629	7.0	1312	0.6617	1.2991
0.7933	8.0	1500	0.6596	1.3467
0.7185	9.0	1687	0.6575	1.3917
0.6489	10.0	1875	0.6550	1.4609
0.586	11.0	2062	0.6531	1.5205
0.5267	12.0	2250	0.6518	1.5616
0.4702	13.0	2437	0.6499	1.6518
0.42	14.0	2625	0.6481	1.7193
0.3811	15.0	2812	0.6459	1.8016
0.3373	16.0	3000	0.6465	1.8556
0.284	17.0	3187	0.6443	1.9407
0.25	18.0	3375	0.6436	1.9760
0.2274	19.0	3562	0.6424	2.1003
0.2033	20.0	3750	0.6425	2.1419
0.1832	21.0	3937	0.6405	2.2005
0.1678	22.0	4125	0.6397	2.2465
0.1505	23.0	4312	0.6400	2.2928
0.1406	24.0	4500	0.6400	2.3691
0.1214	25.0	4687	0.6390	2.4100
0.1128	26.0	4875	0.6382	2.4848
0.1076	27.0	5062	0.6390	2.5469
0.1046	28.0	5250	0.6392	2.5205
0.1007	29.0	5437	0.6382	2.5751
0.0967	30.0	5625	0.6389	2.5988
0.0947	31.0	5812	0.6391	2.6168
0.0914	32.0	6000	0.6393	2.6663
0.0834	33.0	6187	0.6395	2.6627
0.0827	34.0	6375	0.6383	2.6657
0.0818	35.0	6562	0.6390	2.6942
0.0817	36.0	6750	0.6384	2.7361
0.0799	37.0	6937	0.6387	2.7283
0.0784	38.0	7125	0.6380	2.7540
0.0788	39.0	7312	0.6388	2.7975
0.079	40.0	7500	0.6391	2.7849
0.0734	41.0	7687	0.6384	2.8045
0.0735	42.0	7875	0.6389	2.8060
0.073	43.0	8062	0.6387	2.8026
0.0737	44.0	8250	0.6386	2.8394
0.0736	45.0	8437	0.6391	2.8118
0.0724	46.0	8625	0.6388	2.8495
0.0721	47.0	8812	0.6390	2.8442
0.0719	48.0	9000	0.6384	2.8518
0.0687	49.0	9187	0.6385	2.8879
0.0699	49.87	9350	0.6388	2.9168

Framework versions

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