MaRyAm1295/Llama-3.2-3B-KAM

KAM-Llama3.2-3B

This model is a fine-tuned version of meta-llama/Llama-3.2-3B-Instruct on the None dataset.

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: 1
• eval_batch_size: 8
• seed: 3407
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 20
• training_steps: 1200
• mixed_precision_training: Native AMP

Training results

Step    Training Loss

• 50     2.436700
• 100     2.103400
• 150     2.048900
• 200     2.041700
• 250     2.002900
• 300     1.991700
• 350     1.977400
• 400     1.974500
• 450     1.945000
• 500     1.951100
• 550     1.950700
• 600     1.943000
• 650     1.927900
• 700     1.920900
• 750     1.903400
• 800     1.896000
• 850     1.910800
• 900     1.904600
• 950     1.918100
• 1000     1.911500
• 1050     1.909100
• 1100     1.928900
• 1150     1.896100
• 1200     1.876700

Framework versions

• PEFT 0.13.2
• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.19.1