lmind_hotpot_train8000_eval7405_v1_reciteonly_qa_meta-llama_Llama-2-7b-hf_3e-5_lora2

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the tyzhu/lmind_hotpot_train8000_eval7405_v1_reciteonly_qa dataset. It achieves the following results on the evaluation set:

• Loss: 1.3388
• Accuracy: 0.7063

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: 3e-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
1.0715	1.0	250	1.0406	0.7165
1.0108	2.0	500	1.0269	0.7173
0.9956	3.0	750	1.0210	0.7177
0.9652	4.0	1000	1.0227	0.7178
0.9455	5.0	1250	1.0289	0.7172
0.9123	6.0	1500	1.0376	0.7168
0.8831	7.0	1750	1.0475	0.7160
0.8555	8.0	2000	1.0603	0.7153
0.8332	9.0	2250	1.0759	0.7145
0.8103	10.0	2500	1.0939	0.7137
0.7757	11.0	2750	1.1142	0.7130
0.7511	12.0	3000	1.1356	0.7120
0.7305	13.0	3250	1.1520	0.7116
0.7073	14.0	3500	1.1791	0.7104
0.6815	15.0	3750	1.2103	0.7098
0.6602	16.0	4000	1.2371	0.7091
0.6318	17.0	4250	1.2532	0.7084
0.5986	18.0	4500	1.2990	0.7075
0.5737	19.0	4750	1.3180	0.7070
0.5514	20.0	5000	1.3388	0.7063

Framework versions

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