File size: 1,103 Bytes
36ba107 ea3c62a 36ba107 ea3c62a 36ba107 ea3c62a 36ba107 ea3c62a 36ba107 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 |
import ctranslate2
from transformers import AutoTokenizer
import gradio as gr
def generate_prompt(history):
prompt = ""
for chain in history[:-1]:
prompt += f"<human>: {chain[0]}\n<bot>: {chain[1]}\n"
prompt += f"<human>: {history[-1][0]}\n<bot>:"
tokens = tokenizer.convert_ids_to_tokens(tokenizer.encode(prompt))
return tokens
def generate(question):
history = [[question, ""]]
tokens = generate_prompt(history)
results = translator.translate_batch(
[tokens],
beam_size=1,
max_decoding_length = 256,
repetition_penalty = 1.8,
)
answer = tokenizer.convert_tokens_to_string(results[0].hypotheses[0])
return answer
translator = ctranslate2.Translator("model")
tokenizer = AutoTokenizer.from_pretrained("DKYoon/mt5-xl-lm-adapt")
end_token = "</s>"
end_token_id = tokenizer.encode(end_token)[0]
demo = gr.Interface(fn=generate, description="Space by @theodotus. Source: https://huggingface.co/spaces/theodotus/pythia-uk", inputs="text", outputs="text")
demo.queue(1)
if __name__ == "__main__":
demo.launch() |