phi3-spin-zephyr-data

This model is a fine-tuned version of microsoft/Phi-3-small-8k-instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1643
• Rewards/real: -4.3165
• Rewards/generated: -36.8197
• Rewards/accuracies: 0.9626
• Rewards/margins: 32.5032
• Logps/generated: -659.6320
• Logps/real: -295.5523
• Logits/generated: -inf
• Logits/real: -inf

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-07
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) 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	Rewards/real	Rewards/generated	Rewards/accuracies	Rewards/margins	Logps/generated	Logps/real	Logits/generated	Logits/real
0.3303	0.32	500	0.2003	-4.8459	-23.8426	0.9371	18.9967	-529.8613	-300.8461	-inf	-inf
0.0933	0.64	1000	0.1598	-4.6590	-34.8525	0.9610	30.1935	-639.9600	-298.9768	-inf	-inf
0.2065	0.96	1500	0.1643	-4.3165	-36.8197	0.9626	32.5032	-659.6320	-295.5523	-inf	-inf

Framework versions

• Transformers 4.37.0
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2