sft-xcomet_xl_xxl-chosen-10lp-shuff-full-tiny2

This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T on the Unbabel/TowerAligned-v0.1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7084
• Nll Loss: 0.7084
• Logps/best: -70.3876
• Rewards/chosen: 3.2966
• Rewards/rejected: 2.8409
• Rewards/accuracies: 0.6940
• Rewards/margins: 0.4558
• Logps/rejected: -69.0139
• Logps/chosen: -70.3876
• Logits/rejected: -1.7471
• Logits/chosen: -1.8751

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: 1e-05
• train_batch_size: 1
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Nll Loss	Logps/best	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.696	0.5317	500	0.7084	0.7084	-70.3876	3.2966	2.8409	0.6940	0.4558	-69.0139	-70.3876	-1.7471	-1.8751

Framework versions

• Transformers 4.43.3
• Pytorch 2.4.0+cu121
• Datasets 2.17.0
• Tokenizers 0.19.1