Compare output of a ML learning model vs a CSV file

I have a machine learning model on Google Colab and i have this code

wrong_english=[
    "I has a dog",
    "They is going to the park",
    "She don't like coffee.",
    "The book belong to him.",
    "He play soccer very well.",
    "My sister and me are going to the party.",
    "I'm not sure who's car is parked outside.",
    "There is many people in the room.",
    "She sings good.",
]

tokenized=tokenizer(
  wrong_english,
  padding="longest",
  max_length=MAX_LENGTH,
  truncation=True,
  return_tensors='tf'
)
out = model.generate(**tokenized, max_length=128)
print(out)


for i in range(len(wrong_english)):
  print(wrong_english[i]+"------------>"+tokenizer.decode(out[i], skip_special_tokens=True))

and the out put is this one

I has a dog------------>I have a dog.
They is going to the park------------>They are going to the park.
She don't like coffee.------------>She doesn't like coffee.
The book belong to him.------------>The book belongs to him.
He play soccer very well.------------>He plays soccer very well.
My sister and me are going to the party.------------>My sister and me are going to the party.
I'm not sure who's car is parked outside.------------>I'm not sure who's car is parked outside.
There is many people in the room.------------>There are many people in the room.
She sings good.------------>She sings good.

also i have a csv file that looks like this one

how can i compare the output of the ML model with every record in the column B of my CSV file and write the word CORRECT or INCORRECT if the values matches?

e.g

I has a dog------------>I have a dog. -> CORRECT
They is going to the park------------>They are going to the park. -> CORRECT
She don't like coffee.------------>She doesn't like coffee. -> CORRECT
The book belong to him.------------>The book belongs to him. -> CORRECT
He play soccer very well.------------>He plays soccer very well. -> CORRECT
My sister and me are going to the party.------------>My sister and me are going to the party. -> CORRECT
I'm not sure who's car is parked outside.------------>I'm not sure who's car is parked outside. -> INCORRECT
There is many people in the room.------------>There are many people in the room. -> CORRECT
She sings good.------------>She sings good. -> CORRECT

# put your model predictions into a list
predicted_correction = [tokenizer.decode(out[i], skip_special_tokens=True) for i in range(len(wrong_english))]

# read your csv
df = pd.read_csv(CSV_PATH)
# add a correct/incorrect empty result column
df.insert(0,"Result","")

for wr, pc in zip(wrong_english, predicted_correction):
    indexes = df[df['Incorrect - input'] == wr].index
    for i in indexes:  # i'm not excluding the possibility of multiple occurences of the incorrect input
        if df.iloc[i]['Correct - expected output'] == pc:
            df.iloc[i]['Result'] = "Correct"
        else:
            df.iloc[i]['Result'] = "Incorrect"