lmind_hotpot_train8000_eval7405_v1_doc_qa_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_hotpot_train8000_eval7405_v1_doc_qa dataset. It achieves the following results on the evaluation set:

• Loss: 2.6477
• Accuracy: 0.5905

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
1.2049	1.0	1089	1.8022	0.5981
1.0839	2.0	2178	1.7775	0.6020
0.9258	3.0	3267	1.8381	0.5995
0.7903	4.0	4357	1.9065	0.5989
0.6724	5.0	5446	2.0495	0.5959
0.56	6.0	6535	2.1465	0.5950
0.4783	7.0	7624	2.2764	0.5917
0.4167	8.0	8714	2.3067	0.5925
0.373	9.0	9803	2.4418	0.5912
0.3369	10.0	10892	2.4273	0.5923
0.3169	11.0	11981	2.4758	0.5918
0.3008	12.0	13071	2.5142	0.5916
0.2874	13.0	14160	2.5254	0.5914
0.2803	14.0	15249	2.5256	0.5913
0.2752	15.0	16338	2.5693	0.5911
0.2706	16.0	17428	2.5905	0.5897
0.2637	17.0	18517	2.6049	0.5913
0.2577	18.0	19606	2.5941	0.5995
0.2489	19.0	20695	2.6460	0.5892
0.2471	20.0	21780	2.6477	0.5905

Framework versions

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