Sample Question - Error when Converting Phi 3.5 to QNN #2200
Description
I am running sample: https://github.com/microsoft/olive-recipes/tree/main/microsoft-Phi-3.5-mini-instruct/QNN
Running this on a Snapdragon device (Lenovo Slim 7x), to convert to a model that will run on the NPU. I have checked SenstencePiece and Tiktoken is installed.
Am i missing something here? I am relative new to Olive so trying to get a model working.
Do i need to login into HuggingFace and download the source model first, and then run this?
I am getting the following error:
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 30.03it/s]
You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama_fast.LlamaTokenizerFast'>. This is expected, and simply means that the legacy (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set legacy=False. This should only be set if you understand what it means, and thoroughly read the reason why this was added as explained in huggingface/transformers#24565 - if you loaded a llama tokenizer from a GGUF file you can ignore this message.
Traceback (most recent call last):
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\convert_slow_tokenizer.py", line 1737, in convert_slow_tokenizer
).converted()
^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\convert_slow_tokenizer.py", line 1631, in converted
tokenizer = self.tokenizer()
^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\convert_slow_tokenizer.py", line 1624, in tokenizer
vocab_scores, merges = self.extract_vocab_merges_from_model(self.vocab_file)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\convert_slow_tokenizer.py", line 1600, in extract_vocab_merges_from_model
bpe_ranks = load_tiktoken_bpe(tiktoken_url)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\tiktoken\load.py", line 162, in load_tiktoken_bpe
contents = read_file_cached(tiktoken_bpe_file, expected_hash)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\tiktoken\load.py", line 52, in read_file_cached
cache_key = hashlib.sha1(blobpath.encode()).hexdigest()
^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'encode'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in run_code
File "C:\Users\PaulBullock\miniconda3\envs\model\Scripts\olive.exe_main.py", line 6, in
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\cli\launcher.py", line 66, in main
service.run()
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\cli\run.py", line 59, in run
workflow_output = olive_run(
^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\workflows\run\run.py", line 178, in run
return run_engine(package_config, run_config)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\workflows\run\run.py", line 139, in run_engine
return engine.run(
^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\engine\engine.py", line 224, in run
run_result = self.run_accelerator(
^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\engine\engine.py", line 306, in run_accelerator
output_footprint = self._run_no_search(input_model_config, input_model_id, accelerator_spec, output_dir)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\engine\engine.py", line 350, in _run_no_search
should_prune, signal, model_ids = self._run_passes(input_model_config, input_model_id, accelerator_spec)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\engine\engine.py", line 641, in _run_passes
model_config, model_id = self._run_pass(
^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\engine\engine.py", line 736, in _run_pass
output_model_config = host.run_pass(p, input_model_config, output_model_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\systems\local.py", line 45, in run_pass
output_model = the_pass.run(model, output_model_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\passes\olive_pass.py", line 242, in run
output_model = self._run_for_config(model, self.config, output_model_path)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\torch\utils_contextlib.py", line 120, in decorate_context
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\passes\pytorch\rotate.py", line 253, in _run_for_config
model.save_metadata(output_model_path)
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\model\handler\mixin\hf.py", line 107, in save_metadata
get_tokenizer(output_dir)
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\common\hf\utils.py", line 205, in get_tokenizer
tokenizer = from_pretrained(AutoTokenizer, model_name_or_path, "tokenizer", **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\olive\common\hf\utils.py", line 94, in from_pretrained
return cls.from_pretrained(get_pretrained_name_or_path(model_name_or_path, mlflow_dir), **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\models\auto\tokenization_auto.py", line 1069, in from_pretrained
return tokenizer_class_fast.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\tokenization_utils_base.py", line 2014, in from_pretrained
return cls._from_pretrained(
^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\tokenization_utils_base.py", line 2260, in _from_pretrained
tokenizer = cls(*init_inputs, **init_kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\models\llama\tokenization_llama_fast.py", line 154, in init
super().init(
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\tokenization_utils_fast.py", line 139, in init
fast_tokenizer = convert_slow_tokenizer(self, from_tiktoken=True)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\PaulBullock\miniconda3\envs\model\Lib\site-packages\transformers\convert_slow_tokenizer.py", line 1739, in convert_slow_tokenizer
raise ValueError(
ValueError: Converting from SentencePiece and Tiktoken failed, if a converter for SentencePiece is available, provide a model path with a SentencePiece tokenizer.model file.Currently available slow->fast converters: ['AlbertTokenizer', 'BartTokenizer', 'BarthezTokenizer', 'BertTokenizer', 'BigBirdTokenizer', 'BlenderbotTokenizer', 'CamembertTokenizer', 'CLIPTokenizer', 'CodeGenTokenizer', 'ConvBertTokenizer', 'DebertaTokenizer', 'DebertaV2Tokenizer', 'DistilBertTokenizer', 'DPRReaderTokenizer', 'DPRQuestionEncoderTokenizer', 'DPRContextEncoderTokenizer', 'ElectraTokenizer', 'FNetTokenizer', 'FunnelTokenizer', 'GPT2Tokenizer', 'HerbertTokenizer', 'LayoutLMTokenizer', 'LayoutLMv2Tokenizer', 'LayoutLMv3Tokenizer', 'LayoutXLMTokenizer', 'LongformerTokenizer', 'LEDTokenizer', 'LxmertTokenizer', 'MarkupLMTokenizer', 'MBartTokenizer', 'MBart50Tokenizer', 'MPNetTokenizer', 'MobileBertTokenizer', 'MvpTokenizer', 'NllbTokenizer', 'OpenAIGPTTokenizer', 'PegasusTokenizer', 'Qwen2Tokenizer', 'RealmTokenizer', 'ReformerTokenizer', 'RemBertTokenizer', 'RetriBertTokenizer', 'RobertaTokenizer', 'RoFormerTokenizer', 'SeamlessM4TTokenizer', 'SqueezeBertTokenizer', 'T5Tokenizer', 'UdopTokenizer', 'WhisperTokenizer', 'XLMRobertaTokenizer', 'XLNetTokenizer', 'SplinterTokenizer', 'XGLMTokenizer', 'LlamaTokenizer', 'CodeLlamaTokenizer', 'GemmaTokenizer', 'Phi3Tokenizer']