File size: 1,008 Bytes
d8c99c6 052012f cc07ecd dc6f6d7 65b75e6 6aa7c2f 65e2225 dc6f6d7 65e2225 052012f 65e2225 363ebf2 65e2225 052012f 65e2225 dc6f6d7 65e2225 |
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 |
import gradio as gr
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM
tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-2")
model = AutoModelForCausalLM.from_pretrained("microsoft/phi-2")
def eval_text(text):
# Encode the input text
text = "Eres un experto en lenguaje claro. Evalúa el texto siguiente y di si es muy claro, claro o poco claro. El texto es este: " + text
input_ids = tokenizer.encode(text, return_tensors="pt")
# Generate text
out = model.generate(
input_ids,
min_length=100,
max_length=750,
eos_token_id=5,
pad_token_id=1,
top_k=10,
top_p=0.0,
no_repeat_ngram_size=5
)
# Decode the generated output
generated_text = list(map(tokenizer.decode, out))[0]
print(generated_text)
return(f"Result: {generation[0]['generated_text']}")
demo = gr.Interface(fn=eval_text, inputs="text", outputs="text", title="microsoft/phi-2")
demo.launch(share=True) |