zephyr-7b-spin-iter2-v0

This model is a fine-tuned version of argilla/zephyr-7b-spin-iter1-v0 on the argilla/10k_prompts_SPIN_iter2_zephyr_top and the argilla/10k_prompts_SPIN_iter1_zephyr_top dataset.

It achieves the following results on the evaluation set:

• Loss: 0.1253
• Rewards/real: -0.5683
• Rewards/generated: -4.9538
• Rewards/accuracies: 0.9479
• Rewards/margins: 4.3854
• Logps/generated: -739.3701
• Logps/real: -278.2851
• Logits/generated: -2.8430
• Logits/real: -2.8375

MT-Bench results

Model	1st Turn Score	2nd Turn Score	Average Score
zephyr-7b-sft-full	6.6625	6.0250	6.34375
zephyr-7b-spin-iter0-v0	6.64375	6.1750	6.409375
zephyr-7b-spin-iter1-v0	6.90625	6.3000	6.603125
zephyr-7b-spin-iter2-v0	7.1375	6.3125	6.725000
zephyr-7b-spin-iter3-v0	7.09375	6.4500	6.771875

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-07
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/real	Rewards/generated	Rewards/accuracies	Rewards/margins	Logps/generated	Logps/real	Logits/generated	Logits/real
5.8769	0.49	25	0.1890	-0.1680	-2.9833	0.9375	2.8153	-719.6649	-274.2817	-2.7940	-2.8382
0.1202	0.97	50	0.1440	-0.4164	-4.2256	0.9479	3.8092	-732.0879	-276.7652	-2.8395	-2.8439
0.0754	1.46	75	0.1298	-0.5468	-4.7565	0.9583	4.2097	-737.3973	-278.0700	-2.8411	-2.8388
0.0621	1.94	100	0.1253	-0.5683	-4.9538	0.9479	4.3854	-739.3701	-278.2851	-2.8430	-2.8375

Framework versions

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