lmind_nq_train6000_eval6489_v1_doc_qa_v3_meta-llama_Llama-2-7b-chat-hf_lora2

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on the tyzhu/lmind_nq_train6000_eval6489_v1_doc_qa_v3 dataset. It achieves the following results on the evaluation set:

• Loss: 1.9008
• Accuracy: 0.6022

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4745	1.0	529	1.3418	0.6122
1.414	2.0	1058	1.3398	0.5886
1.3204	3.0	1587	1.3659	0.6158
1.1963	4.0	2116	1.4242	0.61
1.0807	5.0	2645	1.5381	0.608
0.9652	6.0	3174	1.6063	0.5807
0.8552	7.0	3703	1.6981	0.6037
0.759	8.0	4232	1.7846	0.6042
0.6433	9.0	4761	1.8386	0.6028
0.5475	10.0	5290	1.9008	0.6022

Framework versions

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