final tweaks

app.py

@@ -144,10 +144,10 @@ FIG_DPI = 300
 
 def get_plot(model_name, plot_eager, generate_type):
     df = pd.DataFrame(BENCHMARK_DATA[generate_type][model_name])
-    df["framework"] = ["PyTorch", "TF (Eager Execition)", "TF (XLA)"]
+    df["framework"] = ["PyTorch", "TF (Eager Execution)", "TF (XLA)"]
     df = pd.melt(df, id_vars=["framework"], value_vars=["T4", "3090", "A100"])
     if plot_eager == "No":
-        df = df[df["framework"] != "TF (Eager Execition)"]
+        df = df[df["framework"] != "TF (Eager Execution)"]
 
     g = sns.catplot(
         data=df,
@@ -155,7 +155,7 @@ def get_plot(model_name, plot_eager, generate_type):
         x="variable",
         y="value",
         hue="framework",
-        palette={"PyTorch": "blue", "TF (Eager Execition)": "orange", "TF (XLA)": "red"},
+        palette={"PyTorch": "blue", "TF (Eager Execution)": "orange", "TF (XLA)": "red"},
         alpha=.9,
     )
     g.despine(left=True)
@@ -177,7 +177,7 @@ with demo:
         """
         # TensorFlow XLA Text Generation Benchmark
         Instructions:
-        1. Pick a tab for the type of generation (or other information);
+        1. Pick a tab for the type of generation (or for benchmark information);
         2. Select a model from the dropdown menu;
         3. Optionally omit results from TensorFlow Eager Execution, if you wish to better compare the performance of
         PyTorch to TensorFlow with XLA.
@@ -269,7 +269,7 @@ with demo:
             gr.Dataframe(
                 headers=["Parameter", "Value"],
                 value=[
-                    ["Transformers Version", "4.22.dev0"],
+                    ["Transformers Version", "4.21"],
                     ["TensorFlow Version", "2.9.1"],
                     ["Pytorch Version", "1.11.0"],
                     ["OS", "22.04 LTS (3090) / Debian 10 (other GPUs)"],