Update README.md

README.md

@@ -3,7 +3,19 @@ library_name: transformers
 license: apache-2.0
 ---
 <!-- Provide a quick summary of what the model is/does. -->
-Pretrained Vision Transformer Neural Quantum State on the \\(J1\\) - \\(J2\\) Heinseberg model on a \\(10\times10\\) square lattice.
+Pretrained Vision Transformer Neural Quantum State on the \\(J_1\\) - \\(J_2\\) Heinseberg model on a \\(10\times10\\) square lattice. 
+The frustration ratio is set to \\(J_2/J_1=0.5\\).
+
+|     Revision    | Variational energy | Time per sweep |                           Description                           |
+|:---------------:|:------------------:|:--------------:|:---------------------------------------------------------------:|
+|       main      |    -0.497505103    |      41s       |       Plain ViT with translation invariance among patches       |
+|      symm_t     |     -0.49760546    |      166s      |                 ViT with translational symmetry                 |
+| symm_trxy_ising |  **-0.497676335**  |              | ViT with translational, point group and sz inversion symmetries |
+
+The time per sweep is evaluated on a single A100-40GB GPU.
+
+The model has been trained by distributing the computation over 40 A100-64GB GPUs for about four days.
+
 
 ## How to Get Started with the Model
 
@@ -68,19 +80,13 @@ The expected output is:
 > Mean:  -0.497479875901942  
 > Mean:  -0.49752966071413424  
 
-The symmetrized wavefunction can be also be downloaded using:
+The fully translational invariant wavefunction can be also be downloaded using:
 
 ```python
 wf = FlaxAutoModel.from_pretrained("nqs-models/j1j2_square_10x10", trust_remote_code=True, revision="symm_t")
 ```
 
-## Training Details
-
-### Training Procedure
-
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-
-The model has been trained on 20 A100 GPUs for 10 hours.
+Use `revision="symm_trxy_ising"` for a wavefunction including also the point group and the sz inversion symmetries.
 
 
 #### Training Hyperparameters
@@ -92,9 +98,7 @@ Number of heads: 12
 
 Total number of parameters: 434760
 
-## Citation [optional]
-
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+## Citation
 
 **BibTeX:** https://www.nature.com/articles/s42005-024-01732-4