pythia410m-test-tldr

This model is a fine-tuned version of mnoukhov/pythia410m-sft-tldr on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6766
• Rewards/chosen: 0.0321
• Rewards/rejected: -0.0032
• Rewards/accuracies: 0.7656
• Rewards/margins: 0.0353
• Logps/rejected: -103.6910
• Logps/chosen: -103.6910
• Logps/ref Rejected: -82.9119
• Logps/ref Chosen: -104.3332

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logps/ref Rejected	Logps/ref Chosen
No log	0.5	4	0.6336	0.1889	0.0621	0.7656	0.1268	-100.5552	-100.5552	-82.9119	-104.3332
No log	1.0	8	0.6766	0.0321	-0.0032	0.7656	0.0353	-103.6910	-103.6910	-82.9119	-104.3332

Framework versions

• PEFT 0.10.0
• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1