import pandas as pd

path = "/home/sg666/MDpLM/benchmarks/MLM"

df = pd.read_csv(path + "/mlm_uppercase_results.csv")

all_sequences = df['Original Sequence'].tolist()
seq_len_sum = sum(len(seq) for seq in all_sequences)
ppls = [ppl for ppl in df['Perplexity'].tolist() if ppl != 10000]

ppl_mean = sum(ppls) / len(ppls)
cos_mean = df.loc[:, 'Cosine Similarity'].mean()
hamming_mean = sum(dist for dist in df['Hamming Distance'].tolist()) / seq_len_sum

print(ppl_mean)
print(cos_mean)
print(hamming_mean)