py : add Phi-1.5/Phi-2 tokenizer (ggml-org#9361)

* add phi2 tokenizer

* add phi name to convert_hf_to_gguf_update.py

* make tokenizer_pre consistent; llama.cpp work

Author: daminho

convert_hf_to_gguf.py

@@ -626,6 +626,9 @@ def get_vocab_base_pre(self, tokenizer) -> str:
         if chkhsh == "4e2b24cc4770243d65a2c9ec19770a72f08cffc161adbb73fcbb6b7dd45a0aae":
             # ref: https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct
             res = "exaone"
+        if chkhsh == "fcace8b9cac38ce847670c970cd5892031a753a1ef381abd1d9af00f713da085":
+            # ref: https://huggingface.co/microsoft/phi-2
+            res = "phi-2"
 
         if res is None:
             logger.warning("\n")

convert_hf_to_gguf_update.py

@@ -98,6 +98,7 @@ class TOKENIZER_TYPE(IntEnum):
     {'name': "bloom",          "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/bigscience/bloom", },
     {'name': "gpt3-finnish",   "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/TurkuNLP/gpt3-finnish-small", },
     {"name": "exaone",         "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct", },
+    {"name": "phi-2",          "tokt": TOKENIZER_TYPE.BPE, "repo": "https://huggingface.co/microsoft/phi-2", },
 ]