akkadian-byt5-full-seed13 – AI Model by kirmala | AlphaNeural AI

You can deploy this model and start earning money today!

kirmala/
akkadian-byt5-full-seed13

like

0

text2text-generation

generated_from_trainer

google/byt5-base

text-generation-inference

endpoints_compatible

Views

No views yet

akkadian-byt5-full-seed13

This model is a fine-tuned version of google/byt5-base on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.4318
Chrf++: 41.6602
Bleu: 20.0142
Geo Mean: 28.8755

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: 0.001
train_batch_size: 2
eval_batch_size: 8
seed: 13
gradient_accumulation_steps: 32
total_train_batch_size: 64
optimizer: Use OptimizerNames.ADAFACTOR and the args are: No additional optimizer arguments
lr_scheduler_type: constant_with_warmup
lr_scheduler_warmup_steps: 100
num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Chrf++	Bleu	Geo Mean
28.3780	0.8621	100	0.7473	8.6978	0.4497	1.9777
20.6800	1.7241	200	0.5995	25.7065	9.2412	15.4130
16.3439	2.5862	300	0.5201	31.4327	11.4675	18.9856
13.7035	3.4483	400	0.4580	39.0813	18.4402	26.8453
11.1459	4.3103	500	0.4642	37.5367	17.1915	25.4030
10.2816	5.0	580	0.4318	41.6602	20.0142	28.8755

Framework versions

Transformers 5.10.1
Pytorch 2.10.0+cu128
Datasets 4.8.5
Tokenizers 0.22.2