Instructions to use dzungpham/graphcodebert-code-classification with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use dzungpham/graphcodebert-code-classification with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("dzungpham/graphcodebert-code-classification", dtype="auto") - Notebooks
- Google Colab
- Kaggle
File size: 1,253 Bytes
34959f8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 | TEST: false
model_name: /kaggle/input/models/dzung271828/microsoft-graphcodebert-base/transformers/default/1
output_dir: training/fourier-spectral-norm-classifier/
num_epochs: 5
max_steps: -1
batch_size: 512
learning_rate: 1.0e-06
max_length: 512
num_labels: 2
use_wandb: false
freeze_base: true
loss_type: ce
focal_alpha: 1.0
focal_gamma: 2.0
r_drop_alpha: 6.0
infonce_temperature: 0.07
infonce_weight: 0.5
seed: 42
resume_from_checkpoint: null
save_steps: 500
eval_steps: 500
logging_steps: 5
label_smoothing: 0.5
adversarial_epsilon: 0.5
use_swa: true
swa_start_epoch: 0
swa_lr: 1.0e-06
data_augmentation: true
aug_rename_prob: 0.7
aug_format_prob: 0.7
weight_decay: 0.1
mixup_alpha: 1.0
low_pass_keep_ratio: 0.5
freq_consistency_weight: 0.2
use_mixcode: true
use_fgm: true
fgm_freq: 5
use_r_drop: true
use_freq_consistency_loss: true
use_attn_spectral: false
attn_spectral_weight: 0.1
attn_spectral_cutoff_ratio: 0.25
hidden_dropout_prob: 0.3
attention_probs_dropout_prob: 0.3
classifier_dropout: 0.4
device: cuda
torch_compile: true
cache_dir: ./tokenized_cache
use_swa_actual: true
use_fgm_actual: true
use_r_drop_actual: true
use_mixcode_actual: true
use_attn_spectral_actual: false
use_freq_consistency_loss_actual: true
use_spectral_norm: true
|