petro-twitter-assistant-30ep

This model is a fine-tuned version of flax-community/gpt-2-spanish on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 4.8837

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: 20
• eval_batch_size: 20
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 40
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
3.123	2.3	1000	3.0761
2.8048	4.6	2000	3.0394
2.5904	6.9	3000	3.0743
2.3804	9.2	4000	3.2378
2.1736	11.49	5000	3.4025
1.9736	13.79	6000	3.6284
1.779	16.09	7000	3.9806
1.5993	18.39	8000	4.2559
1.4584	20.69	9000	4.4938
1.3492	22.99	10000	4.6608
1.2701	25.29	11000	4.8302
1.2309	27.59	12000	4.8696
1.2161	29.89	13000	4.8837

Framework versions

• Transformers 4.20.1
• Pytorch 1.12.0
• Datasets 2.1.0
• Tokenizers 0.12.1