zephyr-7b-lora-64-no-quant-6k

This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the updated and the original datasets. It achieves the following results on the evaluation set:

• Loss: 0.5788
• Rewards/chosen: -0.3940
• Rewards/rejected: -0.7469
• Rewards/accuracies: 0.7200
• Rewards/margins: 0.3529
• Logps/rejected: -332.2092
• Logps/chosen: -323.4424
• Logits/rejected: -2.2597
• Logits/chosen: -2.3729

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-06
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 32
• total_train_batch_size: 256
• 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

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.5916	0.42	100	0.6025	-0.2538	-0.5140	0.6940	0.2602	-308.9146	-309.4196	-2.5166	-2.6027
0.5667	0.84	200	0.5788	-0.3940	-0.7469	0.7200	0.3529	-332.2092	-323.4424	-2.2597	-2.3729

Framework versions

• PEFT 0.7.1
• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2