Describe the bug
Hi AutoLabel team,
Amazing project! Currently, I am trying to use autolabel in Vertex AI workbench, I successfully installed autolabel library and I have Gemini pro access. However, the notebook will raise error when I tried to use Gemini. The only difference my situation with documentation is that I don't have API key in workbench since I use Gemini access from project. Could you help me understand how can I use it?
use Gemini config:
"model": {
"provider": "google",
"name": "gemini-pro", # I also tried "gemini-1.0-pro" both raise same error as follows
"params": {}
},
Error Screenshots
ValidationError Traceback (most recent call last)
Cell In[18], line 2
1 # create an aggent for labeling
----> 2 agent = LabelingAgent(config=config)
File /opt/conda/lib/python3.10/site-packages/autolabel/labeler.py:96, in LabelingAgent.init(self, config, cache, example_selector, create_task, console_output, generation_cache, transform_cache)
92 self.config = (
93 config if isinstance(config, AutolabelConfig) else AutolabelConfig(config)
94 )
95 self.task = TaskFactory.from_config(self.config)
---> 96 self.llm: BaseModel = ModelFactory.from_config(
97 self.config, cache=self.generation_cache
98 )
99 score_type = "logprob_average"
100 if self.config.task_type() == TaskType.ATTRIBUTE_EXTRACTION:
File /opt/conda/lib/python3.10/site-packages/autolabel/models/init.py:47, in ModelFactory.from_config(config, cache)
45 try:
46 model_cls = MODEL_REGISTRY[provider]
---> 47 model_obj = model_cls(config=config, cache=cache)
48 # The below ensures that users should based off of the BaseModel
49 # when creating/registering custom models.
50 assert isinstance(
51 model_obj, BaseModel
52 ), f"{model_obj} should inherit from autolabel.models.BaseModel"
File /opt/conda/lib/python3.10/site-packages/autolabel/models/palm.py:68, in PaLMLLM.init(self, config, cache)
66 self.llm = ChatVertexAI(model_name=self.model_name, **self.model_params)
67 else:
---> 68 self.llm = VertexAI(model_name=self.model_name, **self.model_params)
File /opt/conda/lib/python3.10/site-packages/langchain/load/serializable.py:74, in Serializable.init(self, **kwargs)
73 def init(self, **kwargs: Any) -> None:
---> 74 super().init(**kwargs)
75 self._lc_kwargs = kwargs
File /opt/conda/lib/python3.10/site-packages/pydantic/main.py:341, in pydantic.main.BaseModel.init()
ValidationError: 1 validation error for VertexAI
root
Unknown model publishers/google/models/gemini-1.0-pro; {'gs://google-cloud-aiplatform/schema/predict/instance/text_generation_1.0.0.yaml': <class 'vertexai.preview.language_models._PreviewTextGenerationModel'>} (type=value_error)
Additional context
Add any other context about the problem here.
Describe the bug
Hi AutoLabel team,
Amazing project! Currently, I am trying to use autolabel in Vertex AI workbench, I successfully installed autolabel library and I have Gemini pro access. However, the notebook will raise error when I tried to use Gemini. The only difference my situation with documentation is that I don't have API key in workbench since I use Gemini access from project. Could you help me understand how can I use it?
use Gemini config:
"model": {
"provider": "google",
"name": "gemini-pro", # I also tried "gemini-1.0-pro" both raise same error as follows
"params": {}
},
Error Screenshots
ValidationError Traceback (most recent call last)
Cell In[18], line 2
1 # create an aggent for labeling
----> 2 agent = LabelingAgent(config=config)
File /opt/conda/lib/python3.10/site-packages/autolabel/labeler.py:96, in LabelingAgent.init(self, config, cache, example_selector, create_task, console_output, generation_cache, transform_cache)
92 self.config = (
93 config if isinstance(config, AutolabelConfig) else AutolabelConfig(config)
94 )
95 self.task = TaskFactory.from_config(self.config)
---> 96 self.llm: BaseModel = ModelFactory.from_config(
97 self.config, cache=self.generation_cache
98 )
99 score_type = "logprob_average"
100 if self.config.task_type() == TaskType.ATTRIBUTE_EXTRACTION:
File /opt/conda/lib/python3.10/site-packages/autolabel/models/init.py:47, in ModelFactory.from_config(config, cache)
45 try:
46 model_cls = MODEL_REGISTRY[provider]
---> 47 model_obj = model_cls(config=config, cache=cache)
48 # The below ensures that users should based off of the BaseModel
49 # when creating/registering custom models.
50 assert isinstance(
51 model_obj, BaseModel
52 ), f"{model_obj} should inherit from autolabel.models.BaseModel"
File /opt/conda/lib/python3.10/site-packages/autolabel/models/palm.py:68, in PaLMLLM.init(self, config, cache)
66 self.llm = ChatVertexAI(model_name=self.model_name, **self.model_params)
67 else:
---> 68 self.llm = VertexAI(model_name=self.model_name, **self.model_params)
File /opt/conda/lib/python3.10/site-packages/langchain/load/serializable.py:74, in Serializable.init(self, **kwargs)
73 def init(self, **kwargs: Any) -> None:
---> 74 super().init(**kwargs)
75 self._lc_kwargs = kwargs
File /opt/conda/lib/python3.10/site-packages/pydantic/main.py:341, in pydantic.main.BaseModel.init()
ValidationError: 1 validation error for VertexAI
root
Unknown model publishers/google/models/gemini-1.0-pro; {'gs://google-cloud-aiplatform/schema/predict/instance/text_generation_1.0.0.yaml': <class 'vertexai.preview.language_models._PreviewTextGenerationModel'>} (type=value_error)
Additional context
Add any other context about the problem here.