Update openai_gen.py to explain need of prioritization of tokens

src/monitors4codegen/monitor_guided_decoding/openai_gen.py

@@ -39,7 +39,13 @@ def openai_mgd(
     tokens_sort_key = {k:[0, 0] for k in tokenizer.all_token_ids}
 
     # # TODO: Find a way to prioritize tokens to be blacklisted
-    # # 1. The following code uses info about whether has a break char in it
+
+    # # Why prioritize? OpenAI allows applying logit_bias to upto 300 tokens, whereas the typical number of tokens in vocabulary is 50,000.
+    # # Because of this, it is necessary to identify the top 300 tokens, that we think need to be either blacklisted, or whitelisted.
+    # # This prioritization should be done taking into account what violating token is the model likely to predict in the next step.
+
+    # # Options for prioritization of tokens:
+    # # 1. The following code uses info about whether the token has a break char in it
     # for token, token_id in tokenizer.vocab_trie.iteritems():
     #     if token[0] in monitor.all_break_chars:
     #         tokens_sort_key[token_id][0] = 0 # ".", ", a"
@@ -164,4 +170,4 @@ def convert_bytesrep_to_bytes(x: str) -> bytes:
         gen_tokens += new_all_tokens[all_tokens.shape[0]:].tolist()
         all_tokens = new_all_tokens
 
-    return gen_tokens, gen_text.decode()
+    return gen_tokens, gen_text.decode()