Spaces:
Running
Running
import gradio as gr | |
from translate import translator_fn | |
def predict(text): | |
result = translator_fn(text) | |
return { | |
"input_text": result.input_text, | |
"input_tokens": result.input_tokens, | |
"n_input": result.n_input, | |
"output_text": result.output_text, | |
"output_tokens": result.output_tokens, | |
"n_output": result.n_output, | |
"output_scores": result.output_scores, | |
"cross_attention": result.cross_attention.tolist(), | |
} | |
gradio_app = gr.Interface( | |
predict, | |
inputs=gr.Text(placeholder="Enter a sentence to translate...", label="Input text"), | |
outputs=[gr.Json(description="Model output", label="Model output")], | |
title="En2Ru Scientific Translator", | |
description="Translate scientific texts from English to Russian", | |
examples=[ | |
[ | |
r"There is no closed form to implement the KL divergence by the definition of (REF ) and (REF ) for " | |
r"Gaussian Mixture Models. Instead, we resort to the Monte Carlo simulation method proposed in [1]}. " | |
r"Then, the KL divergence can be caculated by: \(D_{KL_{MC}}(p||q) =\frac{1}{n} \sum _{i=1}^{n} log(" | |
r"\frac{p(x_i)}{q(x_i)})\) \(D_{KL_{MC}}(q||p) =\frac{1}{n} \sum _{i=1}^{n} log(\frac{q(y_i)}{p(y_i)})\)"], | |
[ | |
r"Almost all currently used classifiers are not intrinsically well-calibrated [1]}, which means their " | |
r"output scores can't be interpreted as probabilities. This is an issue when the model is used for " | |
r"decision making, as a component in a more general probabilistic pipeline, or simply when one needs a " | |
r"quantification of the uncertainty in model's predictions, for example in high risk applications."], | |
[ | |
r"First, with the development of the high-torque electric actuators, such as [1]}, [2]} the robots are " | |
r"becoming more dynamical. These actuators allow them not only to move at high speeds, but also to " | |
r"rapidly create forces and torques to perform dynamic actions, such as running, jumping, etc."], | |
], | |
) | |
if __name__ == "__main__": | |
gradio_app.launch() | |