tonyshelby/qwen2.5_7b_checkpoints

adapter

This model is a fine-tuned version of Qwen/Qwen2.5-7B on the aqua_rat_multiple_choice and the aqua_rat_open_form datasets. It achieves the following results on the evaluation set:

• Loss: 0.3698

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: 0.0002
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• total_eval_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.4525	0.1642	100	0.4681
0.4298	0.3284	200	0.4217
0.4128	0.4926	300	0.4075
0.4247	0.6568	400	0.3982
0.3922	0.8210	500	0.3916
0.3634	0.9852	600	0.3876
0.351	1.1494	700	0.3857
0.3613	1.3136	800	0.3825
0.3655	1.4778	900	0.3792
0.3849	1.6420	1000	0.3762
0.3373	1.8062	1100	0.3736
0.358	1.9704	1200	0.3711
0.3476	2.1346	1300	0.3727
0.3318	2.2989	1400	0.3717
0.3309	2.4631	1500	0.3709
0.3141	2.6273	1600	0.3703
0.3252	2.7915	1700	0.3698
0.3446	2.9557	1800	0.3699

Framework versions

• PEFT 0.12.0
• Transformers 4.45.2
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.20.3