HeshamHaroon's picture
Update app.py
bee3802 verified
raw
history blame
980 Bytes
from gradio.inputs import Textbox
from gradio.outputs import Textbox, Table
from gradio import Interface
import aranizer
# Load your tokenizers
tokenizers = {
"aranizer_bpe50k": aranizer.aranizer_bpe50k.get_tokenizer(),
"aranizer_bpe64k": aranizer.aranizer_bpe64k.get_tokenizer(),
"aranizer_sp32k": aranizer.aranizer_sp32k.get_tokenizer(),
# Add more tokenizers as needed
}
def compare_tokenizers(text):
results = []
for name, tokenizer in tokenizers.items():
tokens = tokenizer.tokenize(text)
encoded_output = tokenizer.encode(text, add_special_tokens=True)
decoded_text = tokenizer.decode(encoded_output)
results.append((name, tokens, encoded_output, decoded_text))
return results
inputs = Textbox(label="Enter Arabic text")
outputs = Table(label="Results", columns=["Tokenizer", "Tokens", "Encoded Output", "Decoded Text"])
iface = Interface(fn=compare_tokenizers, inputs=inputs, outputs=outputs)
iface.launch()