OpenThinker-7B-Unverified

This model is a fine-tuned version of Qwen/Qwen2.5-7B-Instruct on the OpenThoughts-Unverified-173k dataset.

Please see OpenThinker-7B for more information.

Model Name	Dataset size	AIME24 I/II	AIME25 I	MATH500	GPQA Diamond	LCBv2
OpenThinker-7B	114k	31.3	30.7	84.4	38.9	41.8
OpenThinker-7B-Unverified	173k	34	29.33	83	39.4	43.8
OpenThinker-32B	114k	66.7	53.3	90.6	61.6	68.9
OpenThinker-32B-Unverified	173k	60.7	44	90	60.6	69.2

Intended uses & limitations

Apache 2.0

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 16
• gradient_accumulation_steps: 6
• total_train_batch_size: 96
• total_eval_batch_size: 128
• 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: 3.0

Framework versions

• Transformers 4.46.1
• Pytorch 2.3.0
• Datasets 3.1.0
• Tokenizers 0.20.3