Qwen2.5-0.5B-Open-R1-Distill-FactThink-SFT

This model is a fine-tuned version of jdqqjr/Qwen2.5-0.5B-Open-R1-Distill on the cot_fact_think_sft_train dataset.

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: 0.0001
  • train_batch_size: 2
  • eval_batch_size: 8
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 4
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 32
  • total_eval_batch_size: 32
  • optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • num_epochs: 6.0
  • mixed_precision_training: Native AMP

Training results

Framework versions

  • Transformers 4.47.0
  • Pytorch 2.3.1+cu121
  • Datasets 2.20.0
  • Tokenizers 0.21.0
Downloads last month
38
Safetensors
Model size
494M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for jdqqjr/Qwen2.5-0.5B-DS-R1-Distill-FactThink-Boxed

Base model

Qwen/Qwen2.5-0.5B
Finetuned
(3)
this model