Hugging Face's logo

fledor
/
xnli_mbert_cda_qwen_multilingual

Safetensors
bert
Model card Files
xet
Community
YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)

XNLI CDA Model with Qwen

This model was trained on the XNLI dataset using Counterfactual Data Augmentation (CDA) with counterfactuals generated by Qwen.

Training Parameters

  • Dataset: XNLI
  • Mode: CDA
  • Selection Model: Qwen
  • Selection Method: Random
  • Train Size: 2400 examples
  • Epochs: 8
  • Batch Size: 24
  • Effective Batch Size: 96 (batch_size * gradient_accumulation_steps)
  • Learning Rate: 3e-05
  • Patience: 4
  • Max Length: 256
  • Gradient Accumulation Steps: 4
  • Warmup Ratio: 0.1
  • Weight Decay: 0.01
  • Optimizer: AdamW
  • Scheduler: cosine_with_warmup
  • Random Seed: 42

Performance

  • Overall Accuracy: 66.13%
  • Overall Loss: 0.0137

Language-Specific Performance

  • English (EN): 73.45%
  • German (DE): 68.42%
  • Arabic (AR): 64.89%
  • Spanish (ES): 69.94%
  • Hindi (HI): 62.32%
  • Swahili (SW): 57.74%

Model Information

  • Base Model: bert-base-multilingual-cased
  • Task: Natural Language Inference
  • Languages: 6 languages (EN, DE, AR, ES, HI, SW)
Downloads last month
5
Safetensors
Model size
0.2B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Collection including fledor/xnli_mbert_cda_qwen_multilingual