Update README.md

Author: mukhal

README.md

@@ -1,5 +1,4 @@
 # Discriminator-Guided Chain-of-Thought Reasoning 
-[[**Paper**]](https://arxiv.org/abs/2305.14934), [[**Website**]](https://mukhal.github.io/grace/)
 
 ![image](https://github.com/mukhal/grace-decoding/assets/5109053/cdb93474-1613-47d8-9bf4-be2ae3086979)
 
@@ -30,6 +29,8 @@ WANDB_MODE=disabled python sample_negative_solutions.py --in_file data/$TASK/tra
 All parameters are self-explanatory, but `--sample_calc` means we will use calculator sampling. That is whenever an operation such as `<< 4 + 5=9 >>` is generated, we will invoke a calculator module to compute the result. 
 
 ### Steps 2 and 3: Alignment and Discriminator Training
+![image](https://github.com/mukhal/grace/assets/5109053/ebbefdc2-0861-4fbc-ad0f-43316741bf58)
+
 Now we want to train a FLAN-T5 encoder as a discriminator over the sampled solutions. 
 ```
 accelerate launch  --mixed_precision=bf16  --num_processes=$GPUS_PER_NODE train_discriminator.py  --task gsm8k \