pixeltable 0.4.0rc2__tar.gz → 0.4.0rc3__tar.gz

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: pixeltable
-Version: 0.4.0rc2
+Version: 0.4.0rc3
 Summary: AI Data Infrastructure: Declarative, Multimodal, and Incremental
 License: Apache-2.0
 Keywords: data-science,machine-learning,database,ai,computer-vision,chatbot,ml,artificial-intelligence,feature-engineering,multimodal,mlops,feature-store,vector-database,llm,genai

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/__version__.py

@@ -1,3 +1,3 @@
 # These version placeholders will be replaced during build.
-__version__ = '0.4.0rc2'
-__version_tuple__ = (0, 4, "0rc2")
+__version__ = '0.4.0rc3'
+__version_tuple__ = (0, 4, "0rc3")

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/anthropic.py

@@ -8,7 +8,7 @@ the [Working with Anthropic](https://pixeltable.readme.io/docs/working-with-anth
 import datetime
 import json
 import logging
-from typing import TYPE_CHECKING, Any, Iterable, Optional, TypeVar, Union, cast
+from typing import TYPE_CHECKING, Any, Iterable, Optional, cast
 
 import httpx
 
@@ -73,16 +73,10 @@ async def messages(
     messages: list[dict[str, str]],
     *,
     model: str,
-    max_tokens: int = 1024,
-    metadata: Optional[dict[str, Any]] = None,
-    stop_sequences: Optional[list[str]] = None,
-    system: Optional[str] = None,
-    temperature: Optional[float] = None,
-    tool_choice: Optional[dict] = None,
-    tools: Optional[list[dict]] = None,
-    top_k: Optional[int] = None,
-    top_p: Optional[float] = None,
-    timeout: Optional[float] = None,
+    max_tokens: int,
+    model_kwargs: Optional[dict[str, Any]] = None,
+    tools: Optional[list[dict[str, Any]]] = None,
+    tool_choice: Optional[dict[str, Any]] = None,
 ) -> dict:
     """
     Create a Message.
@@ -101,25 +95,27 @@ async def messages(
     Args:
         messages: Input messages.
         model: The model that will complete your prompt.
-
-    For details on the other parameters, see: <https://docs.anthropic.com/en/api/messages>
+        model_kwargs: Additional keyword args for the Anthropic `messages` API.
+            For details on the available parameters, see: <https://docs.anthropic.com/en/api/messages>
+        tools: An optional list of Pixeltable tools to use for the request.
+        tool_choice: An optional tool choice configuration.
 
     Returns:
         A dictionary containing the response and other metadata.
 
     Examples:
-        Add a computed column that applies the model `claude-3-haiku-20240307`
+        Add a computed column that applies the model `claude-3-5-sonnet-20241022`
         to an existing Pixeltable column `tbl.prompt` of the table `tbl`:
 
         >>> msgs = [{'role': 'user', 'content': tbl.prompt}]
-        ... tbl.add_computed_column(response=messages(msgs, model='claude-3-haiku-20240307'))
+        ... tbl.add_computed_column(response=messages(msgs, model='claude-3-5-sonnet-20241022'))
     """
-
-    # it doesn't look like count_tokens() actually exists in the current version of the library
+    if model_kwargs is None:
+        model_kwargs = {}
 
     if tools is not None:
         # Reformat `tools` into Anthropic format
-        tools = [
+        model_kwargs['tools'] = [
             {
                 'name': tool['name'],
                 'description': tool['description'],
@@ -132,17 +128,16 @@ async def messages(
             for tool in tools
         ]
 
-    tool_choice_: Optional[dict] = None
     if tool_choice is not None:
         if tool_choice['auto']:
-            tool_choice_ = {'type': 'auto'}
+            model_kwargs['tool_choice'] = {'type': 'auto'}
         elif tool_choice['required']:
-            tool_choice_ = {'type': 'any'}
+            model_kwargs['tool_choice'] = {'type': 'any'}
         else:
             assert tool_choice['tool'] is not None
-            tool_choice_ = {'type': 'tool', 'name': tool_choice['tool']}
+            model_kwargs['tool_choice'] = {'type': 'tool', 'name': tool_choice['tool']}
         if not tool_choice['parallel_tool_calls']:
-            tool_choice_['disable_parallel_tool_use'] = True
+            model_kwargs['tool_choice']['disable_parallel_tool_use'] = True
 
     # make sure the pool info exists prior to making the request
     resource_pool_id = f'rate-limits:anthropic:{model}'
@@ -152,20 +147,8 @@ async def messages(
     # TODO: timeouts should be set system-wide and be user-configurable
     from anthropic.types import MessageParam
 
-    # cast(Any, ...): avoid mypy errors
     result = await _anthropic_client().messages.with_raw_response.create(
-        messages=cast(Iterable[MessageParam], messages),
-        model=model,
-        max_tokens=max_tokens,
-        metadata=_opt(cast(Any, metadata)),
-        stop_sequences=_opt(stop_sequences),
-        system=_opt(system),
-        temperature=_opt(cast(Any, temperature)),
-        tools=_opt(cast(Any, tools)),
-        tool_choice=_opt(cast(Any, tool_choice_)),
-        top_k=_opt(top_k),
-        top_p=_opt(top_p),
-        timeout=_opt(timeout),
+        messages=cast(Iterable[MessageParam], messages), model=model, max_tokens=max_tokens, **model_kwargs
     )
 
     requests_limit_str = result.headers.get('anthropic-ratelimit-requests-limit')
@@ -224,15 +207,6 @@ def _anthropic_response_to_pxt_tool_calls(response: dict) -> Optional[dict]:
     return pxt_tool_calls
 
 
-_T = TypeVar('_T')
-
-
-def _opt(arg: _T) -> Union[_T, 'anthropic.NotGiven']:
-    import anthropic
-
-    return arg if arg is not None else anthropic.NOT_GIVEN
-
-
 __all__ = local_public_names(__name__)
 
 

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/deepseek.py

@@ -1,5 +1,5 @@
 import json
-from typing import TYPE_CHECKING, Any, Optional, Union, cast
+from typing import TYPE_CHECKING, Any, Optional
 
 import httpx
 
@@ -7,8 +7,6 @@ import pixeltable as pxt
 from pixeltable import env
 from pixeltable.utils.code import local_public_names
 
-from .openai import _opt
-
 if TYPE_CHECKING:
     import openai
 
@@ -33,17 +31,9 @@ async def chat_completions(
     messages: list,
     *,
     model: str,
-    frequency_penalty: Optional[float] = None,
-    logprobs: Optional[bool] = None,
-    top_logprobs: Optional[int] = None,
-    max_tokens: Optional[int] = None,
-    presence_penalty: Optional[float] = None,
-    response_format: Optional[dict] = None,
-    stop: Optional[list[str]] = None,
-    temperature: Optional[float] = None,
-    tools: Optional[list[dict]] = None,
-    tool_choice: Optional[dict] = None,
-    top_p: Optional[float] = None,
+    model_kwargs: Optional[dict[str, Any]] = None,
+    tools: Optional[list[dict[str, Any]]] = None,
+    tool_choice: Optional[dict[str, Any]] = None,
 ) -> dict:
     """
     Creates a model response for the given chat conversation.
@@ -60,8 +50,10 @@ async def chat_completions(
     Args:
         messages: A list of messages to use for chat completion, as described in the Deepseek API documentation.
         model: The model to use for chat completion.
-
-    For details on the other parameters, see: <https://api-docs.deepseek.com/api/create-chat-completion>
+        model_kwargs: Additional keyword args for the Deepseek `chat/completions` API.
+            For details on the available parameters, see: <https://api-docs.deepseek.com/api/create-chat-completion>
+        tools: An optional list of Pixeltable tools to use for the request.
+        tool_choice: An optional tool choice configuration.
 
     Returns:
         A dictionary containing the response and other metadata.
@@ -76,39 +68,28 @@ async def chat_completions(
             ]
             tbl.add_computed_column(response=chat_completions(messages, model='deepseek-chat'))
     """
+    if model_kwargs is None:
+        model_kwargs = {}
+
     if tools is not None:
-        tools = [{'type': 'function', 'function': tool} for tool in tools]
+        model_kwargs['tools'] = [{'type': 'function', 'function': tool} for tool in tools]
 
-    tool_choice_: Union[str, dict, None] = None
     if tool_choice is not None:
         if tool_choice['auto']:
-            tool_choice_ = 'auto'
+            model_kwargs['tool_choice'] = 'auto'
         elif tool_choice['required']:
-            tool_choice_ = 'required'
+            model_kwargs['tool_choice'] = 'required'
         else:
             assert tool_choice['tool'] is not None
-            tool_choice_ = {'type': 'function', 'function': {'name': tool_choice['tool']}}
+            model_kwargs['tool_choice'] = {'type': 'function', 'function': {'name': tool_choice['tool']}}
 
-    extra_body: Optional[dict[str, Any]] = None
     if tool_choice is not None and not tool_choice['parallel_tool_calls']:
-        extra_body = {'parallel_tool_calls': False}
+        if 'extra_body' not in model_kwargs:
+            model_kwargs['extra_body'] = {}
+        model_kwargs['extra_body']['parallel_tool_calls'] = False
 
-    # cast(Any, ...): avoid mypy errors
     result = await _deepseek_client().chat.completions.with_raw_response.create(
-        messages=messages,
-        model=model,
-        frequency_penalty=_opt(frequency_penalty),
-        logprobs=_opt(logprobs),
-        top_logprobs=_opt(top_logprobs),
-        max_tokens=_opt(max_tokens),
-        presence_penalty=_opt(presence_penalty),
-        response_format=_opt(cast(Any, response_format)),
-        stop=_opt(stop),
-        temperature=_opt(temperature),
-        tools=_opt(cast(Any, tools)),
-        tool_choice=_opt(cast(Any, tool_choice_)),
-        top_p=_opt(top_p),
-        extra_body=extra_body,
+        messages=messages, model=model, **model_kwargs
     )
 
     return json.loads(result.text)

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/fireworks.py

@@ -5,7 +5,7 @@ first `pip install fireworks-ai` and configure your Fireworks AI credentials, as
 the [Working with Fireworks](https://pixeltable.readme.io/docs/working-with-fireworks) tutorial.
 """
 
-from typing import TYPE_CHECKING, Optional
+from typing import TYPE_CHECKING, Any, Optional
 
 import pixeltable as pxt
 from pixeltable import env
@@ -29,14 +29,7 @@ def _fireworks_client() -> 'fireworks.client.Fireworks':
 
 @pxt.udf(resource_pool='request-rate:fireworks')
 async def chat_completions(
-    messages: list[dict[str, str]],
-    *,
-    model: str,
-    max_tokens: Optional[int] = None,
-    top_k: Optional[int] = None,
-    top_p: Optional[float] = None,
-    temperature: Optional[float] = None,
-    request_timeout: Optional[int] = None,
+    messages: list[dict[str, str]], *, model: str, model_kwargs: Optional[dict[str, Any]] = None
 ) -> dict:
     """
     Creates a model response for the given chat conversation.
@@ -55,8 +48,8 @@ async def chat_completions(
     Args:
         messages: A list of messages comprising the conversation so far.
         model: The name of the model to use.
-
-    For details on the other parameters, see: <https://docs.fireworks.ai/api-reference/post-chatcompletions>
+        model_kwargs: Additional keyword args for the Fireworks `chat_completions` API. For details on the available
+            parameters, see: <https://docs.fireworks.ai/api-reference/post-chatcompletions>
 
     Returns:
         A dictionary containing the response and other metadata.
@@ -70,20 +63,18 @@ async def chat_completions(
         ...     response=chat_completions(messages, model='accounts/fireworks/models/mixtral-8x22b-instruct')
         ... )
     """
-    kwargs = {'max_tokens': max_tokens, 'top_k': top_k, 'top_p': top_p, 'temperature': temperature}
-    kwargs_not_none = {k: v for k, v in kwargs.items() if v is not None}
+    if model_kwargs is None:
+        model_kwargs = {}
 
     # for debugging purposes:
     # res_sync = _fireworks_client().chat.completions.create(model=model, messages=messages, **kwargs_not_none)
     # res_sync_dict = res_sync.dict()
 
-    if request_timeout is None:
-        request_timeout = Config.get().get_int_value('timeout', section='fireworks') or 600
+    if 'request_timeout' not in model_kwargs:
+        model_kwargs['request_timeout'] = Config.get().get_int_value('timeout', section='fireworks') or 600
     # TODO: this timeout doesn't really work, I think it only applies to returning the stream, but not to the timing
     # of the chunks; addressing this would require a timeout for the task running this udf
-    stream = _fireworks_client().chat.completions.acreate(
-        model=model, messages=messages, request_timeout=request_timeout, **kwargs_not_none
-    )
+    stream = _fireworks_client().chat.completions.acreate(model=model, messages=messages, **model_kwargs)
     chunks = []
     async for chunk in stream:
         chunks.append(chunk)

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/gemini.py

@@ -53,8 +53,8 @@ async def generate_content(
         config: Configuration for generation, corresponding to keyword arguments of
             `genai.types.GenerateContentConfig`. For details on the parameters, see:
             <https://googleapis.github.io/python-genai/genai.html#module-genai.types>
-        tools: Optional list of Pixeltable tools to use. It is also possible to specify tools manually via the
-            `config.tools` parameter, but at most one of `config.tools` or `tools` may be used.
+        tools: An optional list of Pixeltable tools to use. It is also possible to specify tools manually via the
+            `config['tools']` parameter, but at most one of `config['tools']` or `tools` may be used.
 
     Returns:
         A dictionary containing the response and other metadata.

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/llama_cpp.py

@@ -17,7 +17,7 @@ def create_chat_completion(
     model_path: Optional[str] = None,
     repo_id: Optional[str] = None,
     repo_filename: Optional[str] = None,
-    args: Optional[dict[str, Any]] = None,
+    model_kwargs: Optional[dict[str, Any]] = None,
 ) -> dict:
     """
     Generate a chat completion from a list of messages.
@@ -35,14 +35,14 @@ def create_chat_completion(
         repo_id: The Hugging Face model repo id (if using a pretrained model).
         repo_filename: A filename or glob pattern to match the model file in the repo (optional, if using a
             pretrained model).
-        args: Additional arguments to pass to the `create_chat_completions` call, such as `max_tokens`, `temperature`,
-            `top_p`, and `top_k`. For details, see the
+        model_kwargs: Additional keyword args for the llama_cpp `create_chat_completions` API, such as `max_tokens`,
+            `temperature`, `top_p`, and `top_k`. For details, see the
             [llama_cpp create_chat_completions documentation](https://llama-cpp-python.readthedocs.io/en/latest/api-reference/#llama_cpp.Llama.create_chat_completion).
     """
     Env.get().require_package('llama_cpp', min_version=[0, 3, 1])
 
-    if args is None:
-        args = {}
+    if model_kwargs is None:
+        model_kwargs = {}
 
     if (model_path is None) == (repo_id is None):
         raise excs.Error('Exactly one of `model_path` or `repo_id` must be provided.')
@@ -56,7 +56,7 @@ def create_chat_completion(
     else:
         Env.get().require_package('huggingface_hub')
         llm = _lookup_pretrained_model(repo_id, repo_filename, n_gpu_layers)
-    return llm.create_chat_completion(messages, **args)  # type: ignore
+    return llm.create_chat_completion(messages, **model_kwargs)  # type: ignore
 
 
 def _is_gpu_available() -> bool:

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/mistralai.py

@@ -5,7 +5,7 @@ first `pip install mistralai` and configure your Mistral AI credentials, as desc
 the [Working with Mistral AI](https://pixeltable.readme.io/docs/working-with-mistralai) tutorial.
 """
 
-from typing import TYPE_CHECKING, Optional, TypeVar, Union
+from typing import TYPE_CHECKING, Any, Optional, TypeVar, Union
 
 import numpy as np
 
@@ -32,16 +32,7 @@ def _mistralai_client() -> 'mistralai.Mistral':
 
 @pxt.udf(resource_pool='request-rate:mistral')
 async def chat_completions(
-    messages: list[dict[str, str]],
-    *,
-    model: str,
-    temperature: Optional[float] = 0.7,
-    top_p: Optional[float] = 1.0,
-    max_tokens: Optional[int] = None,
-    stop: Optional[list[str]] = None,
-    random_seed: Optional[int] = None,
-    response_format: Optional[dict] = None,
-    safe_prompt: Optional[bool] = False,
+    messages: list[dict[str, str]], *, model: str, model_kwargs: Optional[dict[str, Any]] = None
 ) -> dict:
     """
     Chat Completion API.
@@ -60,6 +51,8 @@ async def chat_completions(
     Args:
         messages: The prompt(s) to generate completions for.
         model: ID of the model to use. (See overview here: <https://docs.mistral.ai/getting-started/models/>)
+        model_kwargs: Additional keyword args for the Mistral `chat/completions` API.
+            For details on the available parameters, see: <https://docs.mistral.ai/api/#tag/chat>
 
     For details on the other parameters, see: <https://docs.mistral.ai/api/#tag/chat>
 
@@ -73,34 +66,20 @@ async def chat_completions(
         >>> messages = [{'role': 'user', 'content': tbl.prompt}]
         ... tbl.add_computed_column(response=completions(messages, model='mistral-latest-small'))
     """
+    if model_kwargs is None:
+        model_kwargs = {}
+
     Env.get().require_package('mistralai')
     result = await _mistralai_client().chat.complete_async(
         messages=messages,  # type: ignore[arg-type]
         model=model,
-        temperature=temperature,
-        top_p=top_p,
-        max_tokens=_opt(max_tokens),
-        stop=stop,
-        random_seed=_opt(random_seed),
-        response_format=response_format,  # type: ignore[arg-type]
-        safe_prompt=safe_prompt,
+        **model_kwargs,
     )
     return result.dict()
 
 
 @pxt.udf(resource_pool='request-rate:mistral')
-async def fim_completions(
-    prompt: str,
-    *,
-    model: str,
-    temperature: Optional[float] = 0.7,
-    top_p: Optional[float] = 1.0,
-    max_tokens: Optional[int] = None,
-    min_tokens: Optional[int] = None,
-    stop: Optional[list[str]] = None,
-    random_seed: Optional[int] = None,
-    suffix: Optional[str] = None,
-) -> dict:
+async def fim_completions(prompt: str, *, model: str, model_kwargs: Optional[dict[str, Any]] = None) -> dict:
     """
     Fill-in-the-middle Completion API.
 
@@ -118,6 +97,8 @@ async def fim_completions(
     Args:
         prompt: The text/code to complete.
         model: ID of the model to use. (See overview here: <https://docs.mistral.ai/getting-started/models/>)
+        model_kwargs: Additional keyword args for the Mistral `fim/completions` API.
+            For details on the available parameters, see: <https://docs.mistral.ai/api/#tag/fim>
 
     For details on the other parameters, see: <https://docs.mistral.ai/api/#tag/fim>
 
@@ -130,18 +111,11 @@ async def fim_completions(
 
         >>> tbl.add_computed_column(response=completions(tbl.prompt, model='codestral-latest'))
     """
+    if model_kwargs is None:
+        model_kwargs = {}
+
     Env.get().require_package('mistralai')
-    result = await _mistralai_client().fim.complete_async(
-        prompt=prompt,
-        model=model,
-        temperature=temperature,
-        top_p=top_p,
-        max_tokens=_opt(max_tokens),
-        min_tokens=_opt(min_tokens),
-        stop=stop,
-        random_seed=_opt(random_seed),
-        suffix=_opt(suffix),
-    )
+    result = await _mistralai_client().fim.complete_async(prompt=prompt, model=model, **model_kwargs)
     return result.dict()
 
 

{pixeltable-0.4.0rc2 → pixeltable-0.4.0rc3}/pixeltable/functions/ollama.py

@@ -50,7 +50,7 @@ def generate(
         template: Prompt template to use.
         context: The context parameter returned from a previous call to `generate()`.
         raw: If `True`, no formatting will be applied to the prompt.
-        options: Additional options to pass to the `chat` call, such as `max_tokens`, `temperature`, `top_p`, and
+        options: Additional options for the Ollama `chat` call, such as `max_tokens`, `temperature`, `top_p`, and
             `top_k`. For details, see the
             [Valid Parameters and Values](https://github.com/ollama/ollama/blob/main/docs/modelfile.md#valid-parameters-and-values)
             section of the Ollama documentation.