sengi's picture
Training in progress, step 2000
31b8565 verified
|
raw
history blame
2.59 kB
metadata
library_name: transformers
base_model: maple-research-lab/LLaDOU-v0-Math
tags:
  - generated_from_trainer
model-index:
  - name: LLaDA-planner_balanced
    results: []

LLaDA-planner_balanced

This model is a fine-tuned version of maple-research-lab/LLaDOU-v0-Math on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0

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: 1e-06
  • train_batch_size: 4
  • eval_batch_size: 16
  • seed: 42
  • distributed_type: multi-GPU
  • optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine_warmup_with_min_lr
  • training_steps: 22000

Training results

Training Loss Epoch Step Validation Loss
0.0062 0.0020 1000 0.0
0.0027 0.0039 2000 0.0
0.0045 0.0059 3000 0.0
0.0044 0.0078 4000 0.0
0.0031 0.0098 5000 0.0
0.004 0.0117 6000 0.0
0.0032 0.0137 7000 0.0
0.0043 0.0157 8000 0.0
0.0042 0.0176 9000 0.0
0.0035 0.0196 10000 0.0
0.0043 0.0215 11000 0.0
0.0032 0.0235 12000 0.0
0.0037 0.0254 13000 0.0
0.0034 0.0274 14000 0.0
0.0033 0.0293 15000 0.0
0.0044 0.0313 16000 0.0
0.0011 0.0333 17000 0.0
0.0006 0.0352 18000 0.0
0.0015 0.0372 19000 0.0
0.0018 0.0391 20000 0.0
0.0105 0.0411 21000 0.0
0.0082 0.0430 22000 0.0

Framework versions

  • Transformers 4.57.1
  • Pytorch 2.9.0+cu128
  • Datasets 4.3.0
  • Tokenizers 0.22.1