Update README.md

README.md

@@ -42,6 +42,11 @@ import torch
 tokenizer = AutoTokenizer.from_pretrained("InstaDeepAI/nucleotide-transformer-2.5b-multi-species")
 model = AutoModelForMaskedLM.from_pretrained("InstaDeepAI/nucleotide-transformer-2.5b-multi-species")
 
+# Choose the length to which the input sequences are padded. By default, the 
+# model max length is chosen, but feel free to decrease it as the time taken to 
+# obtain the embeddings increases significantly with it.
+max_length = tokenizer.model_max_length
+
 # Create a dummy dna sequence and tokenize it
 sequences = ["ATTCCGATTCCGATTCCG", "ATTTCTCTCTCTCTCTGAGATCGATCGATCGAT"]
 tokens_ids = tokenizer.batch_encode_plus(sequences, return_tensors="pt", padding="max_length", max_length = max_length)["input_ids"]