openaivec 0.13.6__tar.gz → 0.14.0__tar.gz

{openaivec-0.13.6 → openaivec-0.14.0}/.github/copilot-instructions.md

@@ -13,37 +13,40 @@ This repository-wide guide tells GitHub Copilot how to propose code that fits ou
 
 ## Architecture and roles
 
-- `src/openaivec/proxy.py`
+- `src/openaivec/_proxy.py` (internal)
   - Core batching, deduplication, order preservation, and caching
   - `BatchingMapProxy[S, T]` (sync) / `AsyncBatchingMapProxy[S, T]` (async)
   - The map_func contract is strict: return a list of the same length and order as the inputs
   - Progress bars only in notebook environments via `tqdm.auto`, gated by `show_progress=True`
-- `src/openaivec/responses.py`
+- `src/openaivec/_responses.py` (internal)
   - Batched wrapper over OpenAI Responses JSON-mode API
   - `BatchResponses` / `AsyncBatchResponses` use the proxy internally
   - Retries via `backoff`/`backoff_async` for transient errors (RateLimit, 5xx)
   - Reasoning models (o1/o3 family) must use `temperature=None`; helpful guidance on errors
-- `src/openaivec/embeddings.py`
+- `src/openaivec/_embeddings.py` (internal)
   - Batched embeddings (sync/async)
-- `src/openaivec/pandas_ext.py`
+- `src/openaivec/pandas_ext.py` (public)
   - `Series.ai` / `Series.aio` entry points for responses/embeddings
-  - Uses DI container (`provider.CONTAINER`) to get client and model names
+  - Uses DI container (`_provider.CONTAINER`) to get client and model names
   - Supports batch size, progress, and cache sharing (`*_with_cache`)
-- `src/openaivec/spark.py`
+- `src/openaivec/spark.py` (public)
   - UDF builders: `responses_udf` / `task_udf` / `embeddings_udf` / `count_tokens_udf` / `split_to_chunks_udf`
   - Per-partition duplicate caching to reduce API calls
   - Pydantic → Spark StructType schema conversion
-- `src/openaivec/provider.py`
+- `src/openaivec/_provider.py` (internal)
   - DI container and automatic OpenAI/Azure OpenAI client provisioning
-  - Warns if Azure base URL isn’t v1 format
-- `src/openaivec/util.py`
+  - Warns if Azure base URL isn't v1 format
+- `src/openaivec/_util.py` (internal)
   - `backoff` / `backoff_async` and `TextChunker`
-- Additional modules from CLAUDE.md
-  - `src/openaivec/di.py`: lightweight DI container
-  - `src/openaivec/log.py`: logging/observe helpers
-  - `src/openaivec/prompt.py`: few-shot prompt building
-  - `src/openaivec/serialize.py`: Pydantic schema (de)serialization
-  - `src/openaivec/task/`: pre-built, structured task library
+- Additional internal modules
+  - `src/openaivec/_di.py`: lightweight DI container
+  - `src/openaivec/_log.py`: logging/observe helpers
+  - `src/openaivec/_prompt.py`: few-shot prompt building
+  - `src/openaivec/_serialize.py`: Pydantic schema (de)serialization
+  - `src/openaivec/_model.py`: task configuration models
+  - `src/openaivec/_optimize.py`: performance optimization
+- `src/openaivec/task/` (public)
+  - Pre-built, structured task library
 
 ## Dev commands (uv)
 
@@ -76,17 +79,17 @@ uv run mkdocs serve
 
 ## API contracts and critical rules
 
-- Proxy (BatchingMapProxy / AsyncBatchingMapProxy)
+- Proxy (`_proxy.py` - BatchingMapProxy / AsyncBatchingMapProxy)
   - map_func must return a list with the same length and order as inputs; on mismatch, release events and raise ValueError
   - Inputs are de-duplicated while preserving first-occurrence order; outputs are restored to the original order
   - Progress is only shown in notebooks when `show_progress=True`
   - Async version enforces `max_concurrency` via `asyncio.Semaphore`
-- Responses
+- Responses (`_responses.py`)
   - Use OpenAI Responses JSON mode (`responses.parse`)
   - For reasoning models (o1/o3 families), you MUST set `temperature=None`; helpful error messaging is built-in
   - Strongly prefer structured outputs with Pydantic models
   - Retries with exponential backoff for RateLimit/5xx
-- Embeddings
+- Embeddings (`_embeddings.py`)
   - Return NumPy float32 arrays
 - pandas extensions
   - `.ai.responses` / `.ai.embeddings` strictly preserve Series index and length
@@ -97,9 +100,9 @@ uv run mkdocs serve
   - Convert Pydantic models to Spark schemas; treat Enum/Literal as strings
   - Reasoning models require `temperature=None`
   - Provide token counting and text chunking helpers
-- Provider/DI and Azure
+- Provider/DI and Azure (`_provider.py` / `_di.py`)
   - Auto-detect OpenAI vs Azure OpenAI from env vars
-  - Azure requires v1 base URL (warn otherwise) and uses deployment name as the “model”
+  - Azure requires v1 base URL (warn otherwise) and uses deployment name as the "model"
 
 ## Preferred patterns (Do) and Avoid (Don’t)
 
@@ -138,6 +141,38 @@ Don’t
 - Use `asyncio.run` in async tests (mirrors existing tests)
 - Optional integration tests can run with valid API keys; keep unit tests independent of network
 
+## Package Visibility Guidelines (`__all__`)
+
+### Public API Modules
+These modules are part of the public API and have appropriate `__all__` declarations:
+
+- `pandas_ext.py` - Pandas DataFrame/Series extensions with `.ai/.aio` accessors
+- `spark.py` - Apache Spark UDF builders for distributed processing
+- `task/*` - All task modules (NLP, customer support, table operations)
+
+### Internal Modules (underscore-prefixed)
+These modules are for internal use only and have `__all__ = []`:
+
+- `_embeddings.py` - Batch embedding processing (internal implementation)
+- `_model.py` - Task configuration models (internal types)
+- `_prompt.py` - Few-shot prompt building (internal implementation)  
+- `_responses.py` - Batch response processing (internal implementation)
+- `_util.py`, `_serialize.py`, `_log.py`, `_provider.py`, `_proxy.py`, `_di.py`, `_optimize.py` - Internal utilities
+
+### Main Package API
+Users access core functionality through `__init__.py` exports:
+- `BatchResponses`, `AsyncBatchResponses`
+- `BatchEmbeddings`, `AsyncBatchEmbeddings` 
+- `PreparedTask`, `FewShotPromptBuilder`
+
+### `__all__` Best Practices
+
+1. **Public modules**: Include all classes, functions, and constants intended for external use
+2. **Internal modules**: Use `__all__ = []` to explicitly mark as internal-only
+3. **Task modules**: Each task module should export its main classes/functions
+4. **Package `__init__.py`**: Re-export public API from all public modules
+5. **Consistency**: Maintain alphabetical ordering within `__all__` lists
+
 ## Documentation (MkDocs)
 
 - For new developer-facing APIs, update `docs/api/` and consider a short example under `docs/examples/`
@@ -176,7 +211,7 @@ Don’t
 - pandas `.ai` with shared cache
 
   ```python
-  from openaivec.proxy import BatchingMapProxy
+  from openaivec._proxy import BatchingMapProxy
 
   shared = BatchingMapProxy[str, str](batch_size=64)
   df["text"].ai.responses_with_cache("instructions", cache=shared)

{openaivec-0.13.6 → openaivec-0.14.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: openaivec
-Version: 0.13.6
+Version: 0.14.0
 Summary: Generative mutation for tabular calculation
 Project-URL: Homepage, https://microsoft.github.io/openaivec/
 Project-URL: Repository, https://github.com/microsoft/openaivec
@@ -514,7 +514,7 @@ return rendered prompt with XML format.
 Here is an example:
 
 ```python
-from openaivec.prompt import FewShotPromptBuilder
+from openaivec import FewShotPromptBuilder
 
 prompt: str = (
     FewShotPromptBuilder()
@@ -577,7 +577,7 @@ Here is an example:
 
 ```python
 from openai import OpenAI
-from openaivec.prompt import FewShotPromptBuilder
+from openaivec import FewShotPromptBuilder
 
 client = OpenAI(...)
 model_name = "<your-model-name>"

{openaivec-0.13.6 → openaivec-0.14.0}/README.md

@@ -488,7 +488,7 @@ return rendered prompt with XML format.
 Here is an example:
 
 ```python
-from openaivec.prompt import FewShotPromptBuilder
+from openaivec import FewShotPromptBuilder
 
 prompt: str = (
     FewShotPromptBuilder()
@@ -551,7 +551,7 @@ Here is an example:
 
 ```python
 from openai import OpenAI
-from openaivec.prompt import FewShotPromptBuilder
+from openaivec import FewShotPromptBuilder
 
 client = OpenAI(...)
 model_name = "<your-model-name>"

openaivec-0.14.0/docs/api/main.md

@@ -0,0 +1,118 @@
+# Main Package API
+
+The main `openaivec` package provides the core classes for AI-powered data processing.
+
+## Core Classes
+
+All core functionality is accessible through the main package imports:
+
+::: openaivec.BatchResponses
+    options:
+      members:
+        - of
+        - of_task
+        - parse
+
+::: openaivec.AsyncBatchResponses
+    options:
+      members:
+        - of
+        - of_task
+        - parse
+
+::: openaivec.BatchEmbeddings
+    options:
+      members:
+        - of
+        - create
+
+::: openaivec.AsyncBatchEmbeddings
+    options:
+      members:
+        - of
+        - create
+
+## Task Configuration
+
+::: openaivec.PreparedTask
+
+## Prompt Building
+
+::: openaivec.FewShotPromptBuilder
+    options:
+      members:
+        - purpose
+        - caution
+        - example
+        - improve
+        - build
+        - build_json
+        - get_object
+
+## Usage Examples
+
+### Basic Batch Processing
+
+```python
+from openaivec import BatchResponses
+from openai import OpenAI
+
+# Create batch client
+client = BatchResponses.of(
+    client=OpenAI(),
+    model_name="gpt-4.1-mini"
+)
+
+# Process multiple inputs
+results = client.parse([
+    "Translate 'hello' to French",
+    "What is 2+2?",
+    "Name three colors"
+])
+```
+
+### Structured Outputs with Tasks
+
+```python
+from openaivec import BatchResponses, PreparedTask
+from openai import OpenAI
+from pydantic import BaseModel
+
+class Sentiment(BaseModel):
+    sentiment: str
+    confidence: float
+
+task = PreparedTask(
+    instructions="Analyze sentiment",
+    response_format=Sentiment,
+    temperature=0.0
+)
+
+client = BatchResponses.of_task(
+    client=OpenAI(),
+    model_name="gpt-4.1-mini", 
+    task=task
+)
+
+results = client.parse([
+    "I love this product!",
+    "This is terrible quality"
+])
+```
+
+### Advanced Prompt Building
+
+```python
+from openaivec import FewShotPromptBuilder
+
+prompt = (
+    FewShotPromptBuilder()
+    .purpose("Classify animals by their habitat")
+    .caution("Consider both land and water animals")
+    .example("dolphin", "aquatic")
+    .example("eagle", "aerial") 
+    .example("bear", "terrestrial")
+    .improve()  # AI-powered improvement
+    .build()
+)
+```

{openaivec-0.13.6 → openaivec-0.14.0}/mkdocs.yml

@@ -63,13 +63,9 @@ nav:
       - FAQ Generation: examples/generate_faq.ipynb
       - Token Count and Processing Time: examples/batch_size.ipynb
   - API Reference:
-      - di: api/di.md
+      - Main Package: api/main.md
       - pandas_ext: api/pandas_ext.md
       - spark: api/spark.md
-      - prompt: api/prompt.md
-      - util: api/util.md
-      - responses: api/responses.md
-      - embeddings: api/embeddings.md
       - task: api/task.md
   - Task Domains:
       - Natural Language Processing:

openaivec-0.14.0/src/openaivec/__init__.py

@@ -0,0 +1,13 @@
+from ._embeddings import AsyncBatchEmbeddings, BatchEmbeddings
+from ._model import PreparedTask
+from ._prompt import FewShotPromptBuilder
+from ._responses import AsyncBatchResponses, BatchResponses
+
+__all__ = [
+    "AsyncBatchEmbeddings",
+    "AsyncBatchResponses",
+    "BatchEmbeddings",
+    "BatchResponses",
+    "FewShotPromptBuilder",
+    "PreparedTask",
+]

openaivec-0.13.6/src/openaivec/di.py → openaivec-0.14.0/src/openaivec/_di.py

@@ -2,6 +2,8 @@ from dataclasses import dataclass, field
 from threading import RLock
 from typing import Any, Callable, Dict, Set, Type, TypeVar
 
+__all__ = []
+
 """Simple dependency injection container with singleton lifecycle management.
 
 This module provides a lightweight dependency injection container that manages

openaivec-0.13.6/src/openaivec/embeddings.py → openaivec-0.14.0/src/openaivec/_embeddings.py

@@ -6,9 +6,9 @@ import numpy as np
 from numpy.typing import NDArray
 from openai import AsyncOpenAI, InternalServerError, OpenAI, RateLimitError
 
-from openaivec.log import observe
-from openaivec.proxy import AsyncBatchingMapProxy, BatchingMapProxy
-from openaivec.util import backoff, backoff_async
+from openaivec._log import observe
+from openaivec._proxy import AsyncBatchingMapProxy, BatchingMapProxy
+from openaivec._util import backoff, backoff_async
 
 __all__ = [
     "BatchEmbeddings",

openaivec-0.13.6/src/openaivec/log.py → openaivec-0.14.0/src/openaivec/_log.py

@@ -5,7 +5,7 @@ import uuid
 from logging import Logger
 from typing import Callable
 
-__all__ = ["observe"]
+__all__ = []
 
 
 def observe(logger: Logger):

openaivec-0.13.6/src/openaivec/model.py → openaivec-0.14.0/src/openaivec/_model.py

@@ -1,6 +1,10 @@
 from dataclasses import dataclass
 from typing import Generic, Type, TypeVar
 
+__all__ = [
+    "PreparedTask",
+]
+
 ResponseFormat = TypeVar("ResponseFormat")
 
 

openaivec-0.13.6/src/openaivec/optimize.py → openaivec-0.14.0/src/openaivec/_optimize.py

@@ -5,6 +5,8 @@ from dataclasses import dataclass, field
 from datetime import datetime, timezone
 from typing import List
 
+__all__ = []
+
 
 @dataclass(frozen=True)
 class PerformanceMetric:
@@ -20,7 +22,7 @@ class BatchSizeSuggester:
     min_batch_size: int = 10
     min_duration: float = 30.0
     max_duration: float = 60.0
-    step_ratio: float = 0.1
+    step_ratio: float = 0.2
     sample_size: int = 4
     _history: List[PerformanceMetric] = field(default_factory=list)
     _lock: threading.RLock = field(default_factory=threading.RLock, repr=False)

openaivec-0.13.6/src/openaivec/prompt.py → openaivec-0.14.0/src/openaivec/_prompt.py

@@ -51,8 +51,8 @@ from openai import OpenAI
 from openai.types.responses import ParsedResponse
 from pydantic import BaseModel
 
-from openaivec.model import ResponsesModelName
-from openaivec.provider import CONTAINER
+from openaivec._model import ResponsesModelName
+from openaivec._provider import CONTAINER
 
 __all__ = [
     "FewShotPrompt",

openaivec-0.13.6/src/openaivec/provider.py → openaivec-0.14.0/src/openaivec/_provider.py

@@ -4,8 +4,8 @@ import warnings
 import tiktoken
 from openai import AsyncAzureOpenAI, AsyncOpenAI, AzureOpenAI, OpenAI
 
-from openaivec import di
-from openaivec.model import (
+from openaivec import _di as di
+from openaivec._model import (
     AzureOpenAIAPIKey,
     AzureOpenAIAPIVersion,
     AzureOpenAIBaseURL,
@@ -13,7 +13,9 @@ from openaivec.model import (
     OpenAIAPIKey,
     ResponsesModelName,
 )
-from openaivec.util import TextChunker
+from openaivec._util import TextChunker
+
+__all__ = []
 
 CONTAINER = di.Container()
 

openaivec-0.13.6/src/openaivec/proxy.py → openaivec-0.14.0/src/openaivec/_proxy.py

@@ -4,7 +4,9 @@ from collections.abc import Hashable
 from dataclasses import dataclass, field
 from typing import Any, Awaitable, Callable, Dict, Generic, List, TypeVar
 
-from openaivec.optimize import BatchSizeSuggester
+from openaivec._optimize import BatchSizeSuggester
+
+__all__ = []
 
 S = TypeVar("S", bound=Hashable)
 T = TypeVar("T")

openaivec-0.13.6/src/openaivec/responses.py → openaivec-0.14.0/src/openaivec/_responses.py

@@ -7,10 +7,10 @@ from openai import AsyncOpenAI, BadRequestError, InternalServerError, OpenAI, Ra
 from openai.types.responses import ParsedResponse
 from pydantic import BaseModel
 
-from openaivec.log import observe
-from openaivec.model import PreparedTask, ResponseFormat
-from openaivec.proxy import AsyncBatchingMapProxy, BatchingMapProxy
-from openaivec.util import backoff, backoff_async
+from openaivec._log import observe
+from openaivec._model import PreparedTask, ResponseFormat
+from openaivec._proxy import AsyncBatchingMapProxy, BatchingMapProxy
+from openaivec._util import backoff, backoff_async
 
 __all__ = [
     "BatchResponses",

openaivec-0.13.6/src/openaivec/serialize.py → openaivec-0.14.0/src/openaivec/_serialize.py

@@ -29,7 +29,7 @@ from typing import Any, Dict, List, Literal, Type
 
 from pydantic import BaseModel, Field, create_model
 
-__all__ = ["deserialize_base_model", "serialize_base_model"]
+__all__ = []
 
 
 def serialize_base_model(obj: Type[BaseModel]) -> Dict[str, Any]:

openaivec-0.13.6/src/openaivec/util.py → openaivec-0.14.0/src/openaivec/_util.py

@@ -8,6 +8,8 @@ from typing import Awaitable, Callable, List, Type, TypeVar
 import numpy as np
 import tiktoken
 
+__all__ = []
+
 T = TypeVar("T")
 U = TypeVar("U")
 V = TypeVar("V")

{openaivec-0.13.6 → openaivec-0.14.0}/src/openaivec/pandas_ext.py

@@ -48,13 +48,20 @@ import numpy as np
 import pandas as pd
 import tiktoken
 from openai import AsyncOpenAI, OpenAI
+
+__all__ = [
+    "embeddings_model",
+    "responses_model",
+    "use",
+    "use_async",
+]
 from pydantic import BaseModel
 
-from openaivec.embeddings import AsyncBatchEmbeddings, BatchEmbeddings
-from openaivec.model import EmbeddingsModelName, PreparedTask, ResponseFormat, ResponsesModelName
-from openaivec.provider import CONTAINER, _check_azure_v1_api_url
-from openaivec.proxy import AsyncBatchingMapProxy, BatchingMapProxy
-from openaivec.responses import AsyncBatchResponses, BatchResponses
+from openaivec._embeddings import AsyncBatchEmbeddings, BatchEmbeddings
+from openaivec._model import EmbeddingsModelName, PreparedTask, ResponseFormat, ResponsesModelName
+from openaivec._provider import CONTAINER, _check_azure_v1_api_url
+from openaivec._proxy import AsyncBatchingMapProxy, BatchingMapProxy
+from openaivec._responses import AsyncBatchResponses, BatchResponses
 from openaivec.task.table import FillNaResponse, fillna
 
 __all__ = [
@@ -192,7 +199,7 @@ class OpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.proxy import BatchingMapProxy
+            from openaivec._proxy import BatchingMapProxy
             import numpy as np
 
             # Create a shared cache with custom batch size
@@ -290,8 +297,8 @@ class OpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.model import PreparedTask
-            from openaivec.proxy import BatchingMapProxy
+            from openaivec._model import PreparedTask
+            from openaivec._proxy import BatchingMapProxy
 
             # Create a shared cache with custom batch size
             shared_cache = BatchingMapProxy(batch_size=64)
@@ -323,7 +330,7 @@ class OpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.model import PreparedTask
+            from openaivec._model import PreparedTask
 
             # Assume you have a prepared task for sentiment analysis
             sentiment_task = PreparedTask(...)
@@ -510,7 +517,7 @@ class OpenAIVecDataFrameAccessor:
 
         Example:
             ```python
-            from openaivec.proxy import BatchingMapProxy
+            from openaivec._proxy import BatchingMapProxy
 
             # Create a shared cache with custom batch size
             shared_cache = BatchingMapProxy(batch_size=64)
@@ -607,7 +614,7 @@ class OpenAIVecDataFrameAccessor:
 
         Example:
             ```python
-            from openaivec.model import PreparedTask
+            from openaivec._model import PreparedTask
 
             # Assume you have a prepared task for data analysis
             analysis_task = PreparedTask(...)
@@ -770,7 +777,7 @@ class AsyncOpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.proxy import AsyncBatchingMapProxy
+            from openaivec._proxy import AsyncBatchingMapProxy
 
             # Create a shared cache with custom batch size and concurrency
             shared_cache = AsyncBatchingMapProxy(batch_size=64, max_concurrency=4)
@@ -822,7 +829,7 @@ class AsyncOpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.proxy import AsyncBatchingMapProxy
+            from openaivec._proxy import AsyncBatchingMapProxy
             import numpy as np
 
             # Create a shared cache with custom batch size and concurrency
@@ -878,8 +885,8 @@ class AsyncOpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.model import PreparedTask
-            from openaivec.proxy import AsyncBatchingMapProxy
+            from openaivec._model import PreparedTask
+            from openaivec._proxy import AsyncBatchingMapProxy
 
             # Create a shared cache with custom batch size and concurrency
             shared_cache = AsyncBatchingMapProxy(batch_size=64, max_concurrency=4)
@@ -1027,7 +1034,7 @@ class AsyncOpenAIVecSeriesAccessor:
 
         Example:
             ```python
-            from openaivec.model import PreparedTask
+            from openaivec._model import PreparedTask
 
             # Assume you have a prepared task for sentiment analysis
             sentiment_task = PreparedTask(...)
@@ -1110,7 +1117,7 @@ class AsyncOpenAIVecDataFrameAccessor:
 
         Example:
             ```python
-            from openaivec.proxy import AsyncBatchingMapProxy
+            from openaivec._proxy import AsyncBatchingMapProxy
 
             # Create a shared cache with custom batch size and concurrency
             shared_cache = AsyncBatchingMapProxy(batch_size=64, max_concurrency=4)
@@ -1224,7 +1231,7 @@ class AsyncOpenAIVecDataFrameAccessor:
 
         Example:
             ```python
-            from openaivec.model import PreparedTask
+            from openaivec._model import PreparedTask
 
             # Assume you have a prepared task for data analysis
             analysis_task = PreparedTask(...)

{openaivec-0.13.6 → openaivec-0.14.0}/src/openaivec/spark.py

@@ -12,6 +12,15 @@ improved performance in I/O-bound operations.
 automatically cache duplicate inputs within each partition, significantly reducing
 API calls and costs when processing datasets with overlapping content.
 
+__all__ = [
+    "count_tokens_udf",
+    "embeddings_udf",
+    "responses_udf",
+    "similarity_udf",
+    "split_to_chunks_udf",
+    "task_udf",
+]
+
 ## Setup
 
 First, obtain a Spark session and configure authentication:
@@ -127,10 +136,10 @@ from pyspark.sql.udf import UserDefinedFunction
 from typing_extensions import Literal
 
 from openaivec import pandas_ext
-from openaivec.model import PreparedTask, ResponseFormat
-from openaivec.proxy import AsyncBatchingMapProxy
-from openaivec.serialize import deserialize_base_model, serialize_base_model
-from openaivec.util import TextChunker
+from openaivec._model import PreparedTask, ResponseFormat
+from openaivec._proxy import AsyncBatchingMapProxy
+from openaivec._serialize import deserialize_base_model, serialize_base_model
+from openaivec._util import TextChunker
 
 __all__ = [
     "responses_udf",

{openaivec-0.13.6 → openaivec-0.14.0}/src/openaivec/task/__init__.py

@@ -32,7 +32,7 @@ Specialized tasks for customer service operations:
 ### Quick Start with Default Tasks
 ```python
 from openai import OpenAI
-from openaivec.responses import BatchResponses
+from openaivec._responses import BatchResponses
 from openaivec.task import nlp, customer_support
 
 client = OpenAI()

{openaivec-0.13.6 → openaivec-0.14.0}/src/openaivec/task/customer_support/customer_sentiment.py

@@ -9,7 +9,7 @@ Example:
 
     ```python
     from openai import OpenAI
-    from openaivec.responses import BatchResponses
+    from openaivec._responses import BatchResponses
     from openaivec.task import customer_support
 
     client = OpenAI()
@@ -65,7 +65,7 @@ from typing import List, Literal
 
 from pydantic import BaseModel, Field
 
-from openaivec.model import PreparedTask
+from openaivec._model import PreparedTask
 
 __all__ = ["customer_sentiment"]