try screenshot without html

README.md

@@ -58,6 +58,7 @@ Unlike other distributed inference frameworks, exo does not use a master-worker
 
 Exo supports different [partitioning strategies](exo/topology/partitioning_strategy.py) to split up a model across devices. The default partitioning strategy is [ring memory weighted partitioning](exo/topology/ring_memory_weighted_partitioning_strategy.py). This runs an inference in a ring where each device runs a number of model layers proportional to the memory of the device.
 
+!["A screenshot of exo running 5 nodes](docs/exo-screenshot.png)
 <p>
     <picture>
         <img alt="A screenshot of exo running 5 nodes" src="docs/exo-screenshot.png" width="80%" height="80%">