lmind_nq_train6000_eval6489_v1_reciteonly_qa_v3_Qwen_Qwen1.5-4B_3e-5_lora2

This model is a fine-tuned version of Qwen/Qwen1.5-4B on the tyzhu/lmind_nq_train6000_eval6489_v1_reciteonly_qa_v3 dataset. It achieves the following results on the evaluation set:

• Loss: 2.2508
• Accuracy: 0.5852

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: 1
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 8
• 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
1.8512	0.9973	187	1.7184	0.6032
1.7174	2.0	375	1.7018	0.6049
1.6805	2.9973	562	1.6938	0.6061
1.6412	4.0	750	1.6925	0.6067
1.5834	4.9973	937	1.7047	0.6062
1.5304	6.0	1125	1.7239	0.6056
1.452	6.9973	1312	1.7508	0.6039
1.3847	8.0	1500	1.7711	0.6028
1.3177	8.9973	1687	1.8049	0.6009
1.2747	10.0	1875	1.8298	0.5998
1.2202	10.9973	2062	1.8814	0.5981
1.1589	12.0	2250	1.9311	0.5959
1.1231	12.9973	2437	1.9429	0.5955
1.0624	14.0	2625	1.9969	0.5933
1.0185	14.9973	2812	2.0319	0.5922
0.9718	16.0	3000	2.0798	0.5903
0.9101	16.9973	3187	2.1396	0.5887
0.8606	18.0	3375	2.1882	0.5870
0.8168	18.9973	3562	2.2291	0.5863
0.777	19.9467	3740	2.2508	0.5852

Framework versions

• PEFT 0.5.0
• Transformers 4.40.2
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.19.1