lmind_hotpot_train8000_eval7405_v1_reciteonly_qa_Qwen_Qwen1.5-4B_5e-5_lora2

This model is a fine-tuned version of Qwen/Qwen1.5-4B on the tyzhu/lmind_hotpot_train8000_eval7405_v1_reciteonly_qa dataset. It achieves the following results on the evaluation set:

• Loss: 2.2235
• Accuracy: 0.6607

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: 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
1.4675	1.0	250	1.5235	0.6752
1.435	2.0	500	1.5122	0.6762
1.395	3.0	750	1.5092	0.6761
1.35	4.0	1000	1.5165	0.6761
1.2906	5.0	1250	1.5309	0.6754
1.2411	6.0	1500	1.5509	0.6747
1.1833	7.0	1750	1.5747	0.6737
1.1198	8.0	2000	1.6129	0.6727
1.0498	9.0	2250	1.6407	0.6717
1.0063	10.0	2500	1.6802	0.6706
0.943	11.0	2750	1.7385	0.6691
0.8881	12.0	3000	1.7767	0.6681
0.8176	13.0	3250	1.8362	0.6669
0.7669	14.0	3500	1.8820	0.6659
0.7119	15.0	3750	1.9359	0.6648
0.6564	16.0	4000	2.0029	0.6638
0.6096	17.0	4250	2.0593	0.6631
0.5715	18.0	4500	2.1331	0.6621
0.5293	19.0	4750	2.1593	0.6617
0.4956	20.0	5000	2.2235	0.6607

Framework versions

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