This model is a fine-tuned version of W-61/qwen3-8b-base-sft-ultrachat-4xh200-batch-128 on the HuggingFaceH4/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

Loss: 2432.5354
Rewards/chosen: -0.0070
Rewards/rejected: -0.0123
Rewards/accuracies: 0.6940
Rewards/margins: 0.0053
Logps/rejected: -2.5346
Logps/chosen: -1.7890
Logits/rejected: 1.1293
Logits/chosen: 1.1194

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: 4
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 8
total_train_batch_size: 128
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_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
19697.8125	0.4188	200	2466.3765	-0.0013	-0.0041	0.6920	0.0028	-1.7207	-1.2257	1.2727	1.2777
19439.9469	0.8377	400	2432.5354	-0.0070	-0.0123	0.6940	0.0053	-2.5346	-1.7890	1.1293	1.1194

Framework versions

Transformers 4.51.0
Pytorch 2.3.1+cu121
Datasets 2.21.0
Tokenizers 0.21.4

seanyhan/qwen3-8b-base-ipo-ultrafeedback-4xh200-batch-128-20260422-131855

qwen3-8b-base-ipo-ultrafeedback-4xh200-batch-128-20260422-131855

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

seanyhan/
qwen3-8b-base-ipo-ultrafeedback-4xh200-batch-128-20260422-131855