Upload README.md with huggingface_hub

README.md

@@ -123,9 +123,6 @@ representations serve two downstream purposes:
    can be concatenated to the raw RL state, giving the PPO policy access to
    long-range structural information beyond what the raw int8 array conveys.
 
-Both uses are part of **Proposal B: Transformer-Guided Reinforcement Learning**
-described in the paper.
-
 ---
 
 ## Generation Process