• Model creator: https://huggingface.co/simonycl/
• Original model: https://huggingface.co/simonycl/llama-3.1-8b-instruct-ultrafeedback-agg-judge/

Original model description:

• alignment-handbook
• generated_from_trainer datasets:
• simonycl/llama3.1-ultrafeedback-annotate-judge-5 model-index:
• name: llama-3.1-8b-instruct-agg-judge results: []

---

llama-3.1-8b-instruct-agg-judge

• Loss: 0.6258
• Rewards/chosen: -2.9200
• Rewards/rejected: -3.3645
• Rewards/accuracies: 0.6640
• Rewards/margins: 0.4445
• Logps/rejected: -610.8331
• Logps/chosen: -555.4608
• Logits/rejected: -2.5388
• Logits/chosen: -2.6870

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

• learning_rate: 5e-07
• train_batch_size: 1
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 32
• total_train_batch_size: 128
• 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: 1

Training results

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1