lmind_hotpot_train8000_eval7405_v1_doc_qa_meta-llama_Llama-2-7b-hf_5e-5_lora2

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the tyzhu/lmind_hotpot_train8000_eval7405_v1_doc_qa dataset. It achieves the following results on the evaluation set:

• Loss: 3.1776
• Accuracy: 0.5801

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: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.2112	1.0	1089	1.8415	0.5941
1.1804	2.0	2178	1.8224	0.5972
1.1234	3.0	3267	1.8223	0.5970
1.0733	4.0	4357	1.8320	0.5968
1.0268	5.0	5446	1.8681	0.5968
0.9619	6.0	6535	1.9226	0.5946
0.905	7.0	7624	2.0011	0.5925
0.8459	8.0	8714	2.1081	0.5894
0.7885	9.0	9803	2.2146	0.5881
0.7547	10.0	10892	2.3650	0.5861
0.7093	11.0	11981	2.4565	0.5852
0.6557	12.0	13071	2.5679	0.5823
0.609	13.0	14160	2.5864	0.5835
0.5777	14.0	15249	2.7236	0.5821
0.556	15.0	16338	2.8375	0.5817
0.5114	16.0	17428	2.9108	0.5816
0.4872	17.0	18517	2.9641	0.5807
0.4519	18.0	19606	3.0177	0.5807
0.4173	19.0	20695	3.0646	0.5801
0.3947	20.0	21780	3.1776	0.5801

Framework versions

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