README.md

---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: Hieroglyph-Translator-Using-Gardiner-Codes
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->



# Hieroglyph Gardiner Code Translator Model

This model was created to translate hieroglyphs into english. 

Egyptian Hieroglyphs have been grouped into different classes and given a referencing method called Gardiner Codes using Gardiner Classification.

Using the [Gardiner Codes](https://www.egyptianhieroglyphs.net/gardiners-sign-list/) we can assign meanings to different combinations of hieroglyphs.

To Translate any sequence of hieroglyphs using this model, provide the following input :-

"Translate hieroglyph unicode sequence to English: {Gardiner Codes of the Hieroglyphs}"

Examples :

"Translate hieroglyph unicode sequence to English: A1 B6 F8"

"Translate hieroglyph unicode sequence to English: G4 H9 P3"


## Model description

This model is a fine-tuned version of [t5-small](https://huggingface.co/t5-small) on a custom dataset derived from the [Dictionary of Middle Egyptian](https://archive.org/details/DictionaryOfMiddleEgyptian).

The Inference Api on the hugging face model page doesn't work well, load the model in jupyter notebook using the following code snippet:

      text = "" # add your hieroglyph gardiner code combination in the string
      
      from transformers import AutoTokenizer
      
      tokenizer = AutoTokenizer.from_pretrained("AnushS/hieroglyph_unicode_translator_t5_small")
      inputs = tokenizer(text, return_tensors="pt").input_ids
      
      from transformers import AutoModelForSeq2SeqLM
      
      
      model = AutoModelForSeq2SeqLM.from_pretrained("AnushS/hieroglyph_unicode_translator_t5_small")
      outputs = model.generate(inputs, max_new_tokens=40, do_sample=True, top_k=30, top_p=0.95)
      translated_keywords = str(tokenizer.decode(outputs[0], skip_special_tokens=True))


## Intended uses & limitations

The Model is intended to be used to translate hieroglyphs.
The model does not provide full sentences, it only outputs bits and keywords.


### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2

### Training results

| Training Loss | Epoch | Step | Validation Loss | Gen Len |
|:-------------:|:-----:|:----:|:---------------:|:-------:|
| 5.0665        | 1.0   | 688  | 4.2034          | 6.946   |
| 4.4621        | 2.0   | 1376 | 4.1388          | 6.946   |


### Framework versions

- Transformers 4.27.4
- Pytorch 2.2.0.dev20231113
- Datasets 2.12.0
- Tokenizers 0.13.3