You can deploy this model and start earning money today!

17Lab/
qwen7b-dpo-s11base-dposeed-s12

17Lab/qwen7b-full-sft-s11

other

Views

No views yet

qwen7b_dpo_from_sft_s12

This model is a fine-tuned version of 17Lab/qwen7b-full-sft-s11 on the assimilation_dpo_v1 dataset. It achieves the following results on the evaluation set:

Loss: 0.2094
Rewards/chosen: 0.4219
Rewards/rejected: -1.4433
Rewards/accuracies: 1.0
Rewards/margins: 1.8652
Logps/chosen: -0.7897
Logps/rejected: -17.7186
Logits/chosen: -2.0037
Logits/rejected: -1.9908

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: 4
eval_batch_size: 2
seed: 12
distributed_type: multi-GPU
num_devices: 8
gradient_accumulation_steps: 2
total_train_batch_size: 64
total_eval_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 0.1
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/chosen	Logps/rejected	Logits/chosen	Logits/rejected
0.2031	1.0	25	0.2094	0.4219	-1.4433	1.0	1.8652	-0.7897	-17.7186	-2.0037	-1.9908

Framework versions

PEFT 0.18.1
Transformers 5.6.0
Pytorch 2.7.0+cu128
Datasets 4.0.0
Tokenizers 0.22.2

17Lab/qwen7b-dpo-s11base-dposeed-s12

qwen7b_dpo_from_sft_s12

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

17Lab/
qwen7b-dpo-s11base-dposeed-s12