training_metadata.json

{
  "base_model": "T0KII/MASRIBERT",
  "grandparent_model": "UBC-NLP/MARBERTv2",
  "version": "v2",
  "new_data_sources": [
    "faisalq/EFC-mini",
    "MBZUAI-Paris/Egyptian-SFT-Mixture"
  ],
  "v1_data_sources": [
    "hard",
    "ar_res_reviews",
    "arbml/TEAD",
    "AOC_youm7_comments",
    "RestOf_AOC_youm7_comments",
    "Egyptian_Tweets",
    "TaghreedT",
    "TE_Telecom",
    "TE_Tweets"
  ],
  "block_size": 64,
  "mlm_probability": 0.2,
  "whole_word_mask": true,
  "learning_rate": 2e-05,
  "epochs": 2,
  "effective_batch_size": 128,
  "train_blocks": 1849729,
  "eval_blocks": 18685,
  "final_eval_loss": 2.772667407989502,
  "final_perplexity": 16.00125902152841,
  "intended_downstream_tasks": [
    "sentiment_analysis",
    "emotion_classification",
    "sarcasm_detection"
  ]
}