readme typo again

README.md

@@ -376,7 +376,7 @@ Overall, all models are able to perform to some level of success, but the contex
 
 #### Patent Acceptance Prediction
 
-With access to labeled validation data, the fine-tuned models can be ranked in accuracy. Sample code that does so is provided in `src/val.ipynb` and evaluates 1000 random data samples out of 4000+ samples inside of `src/val.ipynb`.
+With access to labeled validation data, the fine-tuned models can be ranked in accuracy. Sample code that does so is provided in `src/val.ipynb` and evaluates 1000 random data samples out of 4000+ samples.
 
 Overall, both the fine-tuned Abstract model and the Claims model seem to perform very similarly, producing accuracy rates of 72.89% and 72.8% accuracy respectively out of 1000 random samples. The aggregated softmax labeling, assuming equal weighting between the two models, produces 76.2% accuracy. Depending on who you ask, this performance can be discouraging or encouraging.