w3en2g/qa-Qwen2.5-32B-Base

Qwen2.5-32B-Base

This model is a fine-tuned version of Qwen/Qwen2.5-32B on the QA_train_data dataset. It achieves the following results on the evaluation set:

• Loss: 0.9999

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-06
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 16
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• total_eval_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.9373	0.1778	100	0.9672
0.9671	0.3556	200	0.9639
0.9584	0.5333	300	0.9629
0.957	0.7111	400	0.9597
0.9477	0.8889	500	0.9587
0.8552	1.0667	600	0.9710
0.7944	1.2444	700	0.9722
0.7359	1.4222	800	0.9709
0.8494	1.6	900	0.9662
0.8163	1.7778	1000	0.9663
0.8041	1.9556	1100	0.9639
0.6291	2.1333	1200	0.9990
0.6122	2.3111	1300	1.0004
0.6718	2.4889	1400	1.0003
0.6712	2.6667	1500	1.0002
0.6397	2.8444	1600	0.9996

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu124
• Datasets 2.21.0
• Tokenizers 0.20.3