lmind_nq_train6000_eval6489_v1_reciteonly_qa_v3_meta-llama_Llama-2-7b-hf_3e-4_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: 2.2437
• Accuracy: 0.6491

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: 0.0003
• 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: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.2608	1.0	187	1.1864	0.6692
1.0703	2.0	375	1.1833	0.6702
0.8377	3.0	562	1.2356	0.6683
0.5925	4.0	750	1.3472	0.6631
0.3927	5.0	937	1.4937	0.6587
0.2434	6.0	1125	1.6856	0.6541
0.1579	7.0	1312	1.8412	0.6532
0.1164	8.0	1500	1.9470	0.6522
0.0915	9.0	1687	2.0593	0.6519
0.084	10.0	1875	2.1807	0.6507
0.0787	11.0	2062	2.1614	0.6511
0.078	12.0	2250	2.2384	0.6499
0.0778	13.0	2437	2.2370	0.6497
0.0783	14.0	2625	2.2405	0.6507
0.0819	15.0	2812	2.1894	0.6492
0.085	16.0	3000	2.1976	0.6494
0.0801	17.0	3187	2.1847	0.6487
0.0812	18.0	3375	2.2310	0.6487
0.0785	19.0	3562	2.2172	0.6492
0.0771	19.95	3740	2.2437	0.6491

Framework versions

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