fix GUI for default models, update links to doi

Readme.md

@@ -3,7 +3,7 @@ ShimNet is a data-driven AI solution to improve high-resolution nuclear magnetic
 distorted by the inhomogeneous magnetic field (less than optimal shimming). To use it, the experimental training data has to be collected (see **Data collection** below).
 Example data can also be downloaded (see below). 
 
-Paper: [ShimNet: A neural network for post-acquisition improvement of NMR spectra distorted by magnetic-field inhomogeneity](https://pubs.acs.org/doi/full/10.1021/acs.jpcb.5c02632)
+Paper: [ShimNet: A neural network for post-acquisition improvement of NMR spectra distorted by magnetic-field inhomogeneity](https://doi.org/10.1021/acs.jpcb.5c02632)
 
 Web service: [![Open in Hugging Face Spaces](https://huggingface.co/datasets/huggingface/badges/resolve/main/open-in-hf-spaces-md.svg)](https://huggingface.co/spaces/NMR-CeNT-UW/ShimNet)
 

predict-gui.py

@@ -1,3 +1,4 @@
+import os
 import torch
 torch.set_grad_enabled(False)
 import numpy as np
@@ -132,7 +133,7 @@ def process_file(input_file, config_file, weights_file, input_spectrometer_frequ
 # Gradio app
 with gr.Blocks() as app:
     gr.Markdown("# ShimNet Spectra Correction")
-    gr.Markdown("[ShimNet: A neural network for post-acquisition improvement of NMR spectra distorted by magnetic-field inhomogeneity](https://chemrxiv.org/engage/chemrxiv/article-details/67ef86686dde43c90860d315)")
+    gr.Markdown("[ShimNet: A neural network for post-acquisition improvement of NMR spectra distorted by magnetic-field inhomogeneity](https://doi.org/10.1021/acs.jpcb.5c02632)")
     gr.Markdown("Upload your input file, configuration, and weights to process the NMR spectrum.")
 
     with gr.Row():