File size: 2,577 Bytes
8a9003b 2923a84 8a9003b 5f0507e 8a9003b |
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 41 42 43 44 45 46 47 48 49 |
import gradio as gr
import transformers
from transformers import MBartForConditionalGeneration, AutoModelForSeq2SeqLM
from transformers import AlbertTokenizer, AutoTokenizer
tokenizer = AlbertTokenizer.from_pretrained("ai4bharat/MultiIndicWikiBioSS", do_lower_case=False, use_fast=False, keep_accents=True)
# Or use tokenizer = AlbertTokenizer.from_pretrained("ai4bharat/IndicBART-XLSum", do_lower_case=False, use_fast=False, keep_accents=True)
# xlsummodel = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/IndicBART-XLSum")
qgmodel = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/MultiIndicQuestionGenerationSS").eval()
hgmodel = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/MultiIndicHeadlineGenerationSS").eval()
ssmodel = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/MultiIndicSentenceSummarizationSS").eval()
ppmodel = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/MultiIndicParaphraseGenerationSS").eval()
wbmodel = AutoModelForSeq2SeqLM.from_pretrained("ai4bharat/MultiIndicWikiBioSS").eval()
# Some initial mapping
bos_id = tokenizer._convert_token_to_id_with_added_voc("<s>")
eos_id = tokenizer._convert_token_to_id_with_added_voc("</s>")
pad_id = tokenizer._convert_token_to_id_with_added_voc("<pad>")
# To get lang_id use any of ['<2bn>', '<2gu>', '<2hi>', '<2mr>', '<2pa>', '<2ta>', '<2te>']
def greet(choice, lang, input):
if choice == "IndicWikiBio":
model = wbmodel
elif choice == "IndicHeadlineGeneration":
model = hgmodel
elif choice == "IndicParaprasing":
model = ppmodel
elif choice == "IndicSentenceSummarization":
model = ssmodel
elif choice == "IndicQuestionGeneration":
model = qgmodel
inp = tokenizer(input.strip() + " </s> <2" + lang + ">", add_special_tokens=False, return_tensors="pt", padding=True).input_ids
model_output=model.generate(inp, use_cache=True, num_beams=1, max_length=100, min_length=1, early_stopping=True, pad_token_id=pad_id, bos_token_id=bos_id, eos_token_id=eos_id, decoder_start_token_id=tokenizer._convert_token_to_id_with_added_voc("<2"+lang+">"))
# Decode to get output strings
decoded_output=tokenizer.decode(model_output[0], skip_special_tokens=True, clean_up_tokenization_spaces=False)
return decoded_output
iface = gr.Interface(fn=greet, inputs=[gr.inputs.Dropdown("IndicWikiBio", "IndicHeadlineGeneration", "IndicParaprasing", "IndicSentenceSummarization", "IndicQuestionGeneration"), gr.inputs.Dropdown("as","bn", "gu", "hi", "kn", "ml", "mr", "or", "pa", "ta", "te"), "text"], outputs="text")
iface.launch() |