unique_toolkit 0.8.8__py3-none-any.whl → 0.8.10__py3-none-any.whl

unique_toolkit/language_model/functions.py

@@ -20,7 +20,11 @@ from unique_toolkit.language_model import (
     LanguageModelTool,
     LanguageModelToolDescription,
 )
-from unique_toolkit.language_model.infos import LanguageModelName
+from unique_toolkit.language_model.infos import (
+    LanguageModelInfo,
+    LanguageModelName,
+    TemperatureBounds,
+)
 from unique_toolkit.language_model.reference import (
     add_references_to_message,
 )
@@ -303,6 +307,24 @@ def __camelize_keys(data):
     return data
 
 
+def _clamp_temperature(
+    temperature: float, temperature_bounds: TemperatureBounds
+) -> float:
+    temperature = max(temperature_bounds.min_temperature, temperature)
+    temperature = min(temperature_bounds.max_temperature, temperature)
+    return round(temperature, 2)
+
+
+def _prepare_other_options(
+    other_options: dict | None,
+    default_options: dict,
+) -> dict:
+    options = default_options
+    if other_options is not None:
+        options.update(other_options)
+    return options
+
+
 def _prepare_all_completions_params_util(
     messages: LanguageModelMessages | list[ChatCompletionMessageParam],
     model_name: LanguageModelName | str,
@@ -318,6 +340,13 @@ def _prepare_all_completions_params_util(
     list[unique_sdk.Integrated.ChatCompletionRequestMessage],
     dict | None,
 ]:
+    model_info = None
+    if isinstance(model_name, LanguageModelName):
+        model_info = LanguageModelInfo.from_name(model_name)
+        other_options = _prepare_other_options(
+            other_options, model_info.default_options
+        )
+
     if isinstance(messages, LanguageModelMessages):
         options, model, messages_dict, search_context = _prepare_completion_params_util(
             messages=messages,
@@ -341,6 +370,15 @@ def _prepare_all_completions_params_util(
         )
         messages_dict = __camelize_keys(messages.copy())
 
+    if (
+        model_info is not None
+        and model_info.temperature_bounds is not None
+        and "temperature" in options
+    ):
+        options["temperature"] = _clamp_temperature(
+            temperature, model_info.temperature_bounds
+        )
+
     integrated_messages = cast(
         "list[unique_sdk.Integrated.ChatCompletionRequestMessage]",
         messages_dict,

unique_toolkit/language_model/infos.py

@@ -1,6 +1,6 @@
 from datetime import date
 from enum import StrEnum
-from typing import ClassVar, Optional, Self
+from typing import Any, ClassVar, Optional, Self
 
 from pydantic import BaseModel
 from pydantic.json_schema import SkipJsonSchema
@@ -96,6 +96,11 @@ class ModelCapabilities(StrEnum):
     REASONING = "reasoning"
 
 
+class TemperatureBounds(BaseModel):
+    min_temperature: float
+    max_temperature: float
+
+
 class LanguageModelInfo(BaseModel):
     name: LanguageModelName | str
     version: str
@@ -116,6 +121,10 @@ class LanguageModelInfo(BaseModel):
     deprecated_at: date | SkipJsonSchema[None] = None
     retirement_text: str | SkipJsonSchema[None] = None
 
+    temperature_bounds: TemperatureBounds | None = None
+
+    default_options: dict[str, Any] = {}
+
     @classmethod
     def from_name(cls, model_name: LanguageModelName) -> Self:
         match model_name:
@@ -183,11 +192,19 @@ class LanguageModelInfo(BaseModel):
                         ModelCapabilities.STRUCTURED_OUTPUT,
                         ModelCapabilities.PARALLEL_FUNCTION_CALLING,
                     ],
-                    token_limits=LanguageModelTokenLimits(token_limit_input=272000, token_limit_output=128000),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=272000, token_limit_output=128000
+                    ),
                     info_cutoff_at=date(2024, 10, 24),
                     published_at=date(2025, 8, 7),
                     deprecated_at=date(2026, 8, 7),
                     retirement_at=date(2026, 8, 7),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "minimal",
+                    },
                 )
             case LanguageModelName.AZURE_GPT_5_MINI_2025_0807:
                 return cls(
@@ -201,11 +218,19 @@ class LanguageModelInfo(BaseModel):
                         ModelCapabilities.VISION,
                         ModelCapabilities.STRUCTURED_OUTPUT,
                     ],
-                    token_limits=LanguageModelTokenLimits(token_limit_input=272000, token_limit_output=128000),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=272000, token_limit_output=128000
+                    ),
                     info_cutoff_at=date(2024, 6, 24),
                     published_at=date(2025, 8, 7),
                     deprecated_at=date(2026, 8, 7),
                     retirement_at=date(2026, 8, 7),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "minimal",
+                    },
                 )
             case LanguageModelName.AZURE_GPT_5_NANO_2025_0807:
                 return cls(
@@ -219,19 +244,29 @@ class LanguageModelInfo(BaseModel):
                         ModelCapabilities.VISION,
                         ModelCapabilities.STRUCTURED_OUTPUT,
                     ],
-                    token_limits=LanguageModelTokenLimits(token_limit_input=272000, token_limit_output=128000),
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=272000, token_limit_output=128000
+                    ),
                     info_cutoff_at=date(2024, 5, 31),
                     published_at=date(2025, 8, 7),
                     deprecated_at=date(2026, 8, 7),
                     retirement_at=date(2026, 8, 7),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "minimal",
+                    },
                 )
             case LanguageModelName.AZURE_GPT_5_CHAT_2025_0807:
                 return cls(
                     name=model_name,
                     provider=LanguageModelProvider.AZURE,
                     version="2025-08-07",
-                    encoder_name=EncoderName.O200K_BASE, 
-                    token_limits=LanguageModelTokenLimits(token_limit_input=128000, token_limit_output=16384),
+                    encoder_name=EncoderName.O200K_BASE,
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=128000, token_limit_output=16384
+                    ),
                     info_cutoff_at=date(2024, 10, 24),
                     published_at=date(2025, 8, 7),
                     deprecated_at=date(2026, 8, 7),
@@ -347,6 +382,9 @@ class LanguageModelInfo(BaseModel):
                     ),
                     info_cutoff_at=date(2023, 10, 1),
                     published_at=date(2024, 9, 12),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
                 )
             case LanguageModelName.AZURE_o1_2024_1217:
                 return cls(
@@ -366,6 +404,9 @@ class LanguageModelInfo(BaseModel):
                     ),
                     info_cutoff_at=date(2023, 10, 1),
                     published_at=date(2024, 12, 17),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
                 )
             case LanguageModelName.AZURE_o3_MINI_2025_0131:
                 return cls(
@@ -384,6 +425,9 @@ class LanguageModelInfo(BaseModel):
                     ),
                     info_cutoff_at=date(2023, 10, 1),
                     published_at=date(2025, 1, 31),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
                 )
             case LanguageModelName.AZURE_o3_2025_0416:
                 return cls(
@@ -403,6 +447,9 @@ class LanguageModelInfo(BaseModel):
                     ),
                     info_cutoff_at=date(2024, 5, 31),
                     published_at=date(2025, 4, 16),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
                 )
             case LanguageModelName.AZURE_o4_MINI_2025_0416:
                 return cls(
@@ -422,6 +469,9 @@ class LanguageModelInfo(BaseModel):
                     ),
                     info_cutoff_at=date(2024, 5, 31),
                     published_at=date(2025, 4, 16),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
                 )
             case LanguageModelName.AZURE_GPT_45_PREVIEW_2025_0227:
                 return cls(
@@ -507,7 +557,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-3-7-sonnet",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=128_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=128_000,
                     ),
                     info_cutoff_at=date(2024, 10, 31),
                     published_at=date(2025, 2, 24),
@@ -525,7 +577,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-3-7-sonnet-thinking",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=128_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=128_000,
                     ),
                     info_cutoff_at=date(2024, 10, 31),
                     published_at=date(2025, 2, 24),
@@ -543,7 +597,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-sonnet-4",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=64_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=64_000,
                     ),
                     info_cutoff_at=date(2025, 3, 1),
                     published_at=date(2025, 5, 1),
@@ -561,7 +617,9 @@ class LanguageModelInfo(BaseModel):
                     version="claude-opus-4",
                     encoder_name=EncoderName.O200K_BASE,  # TODO: Update encoder with litellm
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=200_000, token_limit_output=32_000
+                        # Input limit is 200_000, we leave 20_000 tokens as buffer due to tokenizer mismatch
+                        token_limit_input=180_000,
+                        token_limit_output=32_000,
                     ),
                     info_cutoff_at=date(2025, 3, 1),
                     published_at=date(2025, 5, 1),

{unique_toolkit-0.8.8.dist-info → unique_toolkit-0.8.10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: unique_toolkit
-Version: 0.8.8
+Version: 0.8.10
 Summary: 
 License: Proprietary
 Author: Martin Fadler
@@ -114,6 +114,13 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [0.8.10] - 2025-08-15
+- Add min and max temperature to `LanguageModelInfo`: temperature will be clamped to the min and max temperature
+- Add default options to `LanguageModelInfo`: These are used by default
+
+## [0.8.9] - 2025-08-15
+- Reduce input token limits for `ANTHROPIC_CLAUDE_3_7_SONNET_THINKING`, `ANTHROPIC_CLAUDE_3_7_SONNET`, `ANTHROPIC_CLAUDE_OPUS_4` and `ANTHROPIC_CLAUDE_SONNET_4` to 180_000 from 200_000
+
 ## [0.8.8] - 2025-08-11
 - Make chat service openai stream response openai compatible 
 - Make `ChatMessage` openai compatible

{unique_toolkit-0.8.8.dist-info → unique_toolkit-0.8.10.dist-info}/RECORD

@@ -54,8 +54,8 @@ unique_toolkit/framework_utilities/utils.py,sha256=JK7g2yMfEx3eMprug26769xqNpS5W
 unique_toolkit/language_model/__init__.py,sha256=lRQyLlbwHbNFf4-0foBU13UGb09lwEeodbVsfsSgaCk,1971
 unique_toolkit/language_model/builder.py,sha256=4OKfwJfj3TrgO1ezc_ewIue6W7BCQ2ZYQXUckWVPPTA,3369
 unique_toolkit/language_model/constants.py,sha256=B-topqW0r83dkC_25DeQfnPk3n53qzIHUCBS7YJ0-1U,119
-unique_toolkit/language_model/functions.py,sha256=qi4KGa4hLUXy0KwkkGmtGBM1IuYfJqmTheSQ95R7pNw,15611
-unique_toolkit/language_model/infos.py,sha256=mnUnbjDQNOIuPS2VE1SsgyKOxdRulo-9Z5k7_S1Q8Cw,37631
+unique_toolkit/language_model/functions.py,sha256=-nWgcscaZvEfuf2ftCQj8hyFY2RynPFO8XgX2gsN924,16665
+unique_toolkit/language_model/infos.py,sha256=7huMpiDhGdWyhFBZx9dVLXdSkONSrxHL1Fr7zTBeX0w,40062
 unique_toolkit/language_model/prompt.py,sha256=JSawaLjQg3VR-E2fK8engFyJnNdk21zaO8pPIodzN4Q,3991
 unique_toolkit/language_model/reference.py,sha256=nkX2VFz-IrUz8yqyc3G5jUMNwrNpxITBrMEKkbqqYoI,8583
 unique_toolkit/language_model/schemas.py,sha256=p95cAwrYmEBupI8f7JZY17aJTKEHvtqF88hMIzS5r_k,16259
@@ -82,7 +82,7 @@ unique_toolkit/tools/utils/execution/execution.py,sha256=vjG2Y6awsGNtlvyQAGCTthQ
 unique_toolkit/tools/utils/source_handling/schema.py,sha256=pvNhtL2daDLpCVIQpfdn6R35GvKmITVLXjZNLAwpgUE,871
 unique_toolkit/tools/utils/source_handling/source_formatting.py,sha256=C7uayNbdkNVJdEARA5CENnHtNY1SU6etlaqbgHNyxaQ,9152
 unique_toolkit/tools/utils/source_handling/tests/test_source_formatting.py,sha256=zu3AJnYH9CMqZPrxKEH3IgI-fM3nlvIBuspJG6W6B18,6978
-unique_toolkit-0.8.8.dist-info/LICENSE,sha256=GlN8wHNdh53xwOPg44URnwag6TEolCjoq3YD_KrWgss,193
-unique_toolkit-0.8.8.dist-info/METADATA,sha256=AUly1ya9VxSfPCq8ivY1kAq4MrClSRQ-fm8_9BZr0oI,26649
-unique_toolkit-0.8.8.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-unique_toolkit-0.8.8.dist-info/RECORD,,
+unique_toolkit-0.8.10.dist-info/LICENSE,sha256=GlN8wHNdh53xwOPg44URnwag6TEolCjoq3YD_KrWgss,193
+unique_toolkit-0.8.10.dist-info/METADATA,sha256=GVFDjq3YzjeBV-w8WTtrqLKSuCYuPTc42lgv7Ev9hWk,27071
+unique_toolkit-0.8.10.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+unique_toolkit-0.8.10.dist-info/RECORD,,