Merge branch 'main' of https://github.com/molmod/psiflow

README.md

@@ -17,7 +17,7 @@ Psiflow is a **modular** and **scalable** library for developing interatomic pot
 Its features include:
 
 - active learning algorithms with enhanced sampling using PLUMED
-- [Weights & Biases](wandb.ai) logging for easy monitoring and analysis
+- [Weights & Biases](https://wandb.ai) logging for easy monitoring and analysis
 - periodic (CP2K) and nonperiodic (PySCF, NWChem) systems
 - efficient GPU-accelerated molecular dynamics using OpenMM
 - the latest equivariant potentials such as MACE and NequIP

configs/hortense.yaml

@@ -29,13 +29,14 @@ ModelTraining:
     exclusive: false
     scheduler_options: "#SBATCH --clusters=dodrio\n#SBATCH --gpus=1\n"
 ReferenceEvaluation:
+  cores_per_worker: 32
   max_walltime: 20
-  cpu_affinity: 'alternating'
+  cpu_affinity: 'none'
   SlurmProvider:
     partition: "cpu_rome"
     account: "2022_069"
     nodes_per_block: 1
-    cores_per_node: 64
+    cores_per_node: 32
     init_blocks: 1
     min_blocks: 0 
     max_blocks: 10 

examples/README.md

@@ -1,5 +1,5 @@
 These examples illustrate some of the core features of psiflow. All execution-side parameters should be defined in a [psiflow configuration file](/configs). Next, simply run the example and pass the configuration file as an argument:
 ```
-python <example.py>                                    # for local execution
-python <example.py> eurohpc_lumi.yaml   # remote execution on LUMI
+python <example.py>             # for local execution
+python <example.py> lumi.yaml   # remote execution on LUMI
 ```