You can deploy this model and start earning money today!

17Lab/
qwen14b-dpo-seed12-r64-s42

Name: qwen14b-dpo-seed12-r64-s42
Availability: InStock

17Lab/qwen14b-strict-json-full-sft-s42

other

Views

No views yet

qwen14b_dpo_from_sft_seed_s12

This model is a fine-tuned version of 17Lab/qwen14b-strict-json-full-sft-s42 on the assimilation_dpo_v1 dataset. It achieves the following results on the evaluation set:

Loss: 0.2515
Rewards/chosen: -0.0479
Rewards/rejected: -1.8479
Rewards/accuracies: 1.0
Rewards/margins: 1.8000
Logps/chosen: -4.6400
Logps/rejected: -36.3794
Logits/chosen: -0.9764
Logits/rejected: -0.9780

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: 1
seed: 12
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 16
total_train_batch_size: 64
total_eval_batch_size: 2
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
0.2437	1.0	25	0.2515	-0.0479	-1.8479	1.0	1.8000	-4.6400	-36.3794	-0.9764	-0.9780

Framework versions

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

17Lab/qwen14b-dpo-seed12-r64-s42

qwen14b_dpo_from_sft_seed_s12

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

17Lab/
qwen14b-dpo-seed12-r64-s42