langroid 0.31.3__py3-none-any.whl → 0.32.0__py3-none-any.whl

langroid/language_models/openai_gpt.py

@@ -67,6 +67,7 @@ if "OLLAMA_HOST" in os.environ:
 else:
     OLLAMA_BASE_URL = "http://localhost:11434/v1"
 
+DEEPSEEK_BASE_URL = "https://api.deepseek.com/v1"
 OPENROUTER_BASE_URL = "https://openrouter.ai/api/v1"
 GEMINI_BASE_URL = "https://generativelanguage.googleapis.com/v1beta/openai"
 GLHF_BASE_URL = "https://glhf.chat/api/openai/v1"
@@ -77,6 +78,10 @@ VLLM_API_KEY = os.environ.get("VLLM_API_KEY", DUMMY_API_KEY)
 LLAMACPP_API_KEY = os.environ.get("LLAMA_API_KEY", DUMMY_API_KEY)
 
 
+class DeepSeekModel(str, Enum):
+    DEEPSEEK = "deepseek/deepseek-chat"
+
+
 class AnthropicModel(str, Enum):
     """Enum for Anthropic models"""
 
@@ -130,6 +135,11 @@ _context_length: Dict[str, int] = {
     AnthropicModel.CLAUDE_3_OPUS: 200_000,
     AnthropicModel.CLAUDE_3_SONNET: 200_000,
     AnthropicModel.CLAUDE_3_HAIKU: 200_000,
+    DeepSeekModel.DEEPSEEK: 64_000,
+    GeminiModel.GEMINI_2_FLASH: 1_000_000,
+    GeminiModel.GEMINI_1_5_FLASH: 1_000_000,
+    GeminiModel.GEMINI_1_5_FLASH_8B: 1_000_000,
+    GeminiModel.GEMINI_1_5_PRO: 2_000_000,
 }
 
 _cost_per_1k_tokens: Dict[str, Tuple[float, float]] = {
@@ -146,6 +156,8 @@ _cost_per_1k_tokens: Dict[str, Tuple[float, float]] = {
     AnthropicModel.CLAUDE_3_OPUS: (0.015, 0.075),
     AnthropicModel.CLAUDE_3_SONNET: (0.003, 0.015),
     AnthropicModel.CLAUDE_3_HAIKU: (0.00025, 0.00125),
+    DeepSeekModel.DEEPSEEK: (0.00014, 0.00028),
+    # Gemini models have complex pricing based on input-len
 }
 
 
@@ -444,6 +456,8 @@ class OpenAIGPT(LanguageModel):
         config = config.copy()
         super().__init__(config)
         self.config: OpenAIGPTConfig = config
+        # save original model name such as `provider/model` before
+        # we strip out the `provider`
         self.chat_model_orig = self.config.chat_model
 
         # Run the first time the model is used
@@ -569,6 +583,7 @@ class OpenAIGPT(LanguageModel):
         self.is_groq = self.config.chat_model.startswith("groq/")
         self.is_cerebras = self.config.chat_model.startswith("cerebras/")
         self.is_gemini = self.is_gemini_model()
+        self.is_deepseek = self.is_deepseek_model()
         self.is_glhf = self.config.chat_model.startswith("glhf/")
         self.is_openrouter = self.config.chat_model.startswith("openrouter/")
 
@@ -609,6 +624,10 @@ class OpenAIGPT(LanguageModel):
                 )
                 self.api_key = os.getenv("OPENROUTER_API_KEY", DUMMY_API_KEY)
                 self.api_base = OPENROUTER_BASE_URL
+            elif self.is_deepseek:
+                self.config.chat_model = self.config.chat_model.replace("deepseek/", "")
+                self.api_base = DEEPSEEK_BASE_URL
+                self.api_key = os.getenv("DEEPSEEK_API_KEY", DUMMY_API_KEY)
 
             self.client = OpenAI(
                 api_key=self.api_key,
@@ -699,6 +718,13 @@ class OpenAIGPT(LanguageModel):
             "gemini/"
         )
 
+    def is_deepseek_model(self) -> bool:
+        deepseek_models = [e.value for e in DeepSeekModel]
+        return (
+            self.chat_model_orig in deepseek_models
+            or self.chat_model_orig.startswith("deepseek/")
+        )
+
     def requires_first_user_message(self) -> bool:
         """
         Does the chat_model require a non-empty first user message?
@@ -711,7 +737,7 @@ class OpenAIGPT(LanguageModel):
         """
         List of params that are not supported by the current model
         """
-        match self.config.chat_model:
+        match self.chat_model_orig:
             case OpenAIChatModel.O1_MINI | OpenAIChatModel.O1_PREVIEW:
                 return ["temperature", "stream"]
             case _:
@@ -764,7 +790,7 @@ class OpenAIGPT(LanguageModel):
         models/endpoints.
         Get it from the dict, otherwise fail-over to general method
         """
-        return _cost_per_1k_tokens.get(self.config.chat_model, super().chat_cost())
+        return _cost_per_1k_tokens.get(self.chat_model_orig, super().chat_cost())
 
     def set_stream(self, stream: bool) -> bool:
         """Enable or disable streaming output from API.

langroid/language_models/utils.py

@@ -53,6 +53,10 @@ def retry_with_exponential_backoff(
                 logger.error(f"OpenAI API request failed with error: {e}.")
                 raise e
 
+            except openai.UnprocessableEntityError as e:
+                logger.error(f"OpenAI API request failed with error: {e}.")
+                raise e
+
             # Retry on specified errors
             except errors as e:
                 # Increment retries

{langroid-0.31.3.dist-info → langroid-0.32.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: langroid
-Version: 0.31.3
+Version: 0.32.0
 Summary: Harness LLMs with Multi-Agent Programming
 License: MIT
 Author: Prasad Chalasani

{langroid-0.31.3.dist-info → langroid-0.32.0.dist-info}/RECORD

@@ -88,12 +88,12 @@ langroid/language_models/azure_openai.py,sha256=zNQzzsERxNestq-hFfQZbvTzK43G2vjR
 langroid/language_models/base.py,sha256=6hXR-bclyPif-BvFbyXevP-gEwiawQAJHX3N1AKNei0,23786
 langroid/language_models/config.py,sha256=9Q8wk5a7RQr8LGMT_0WkpjY8S4ywK06SalVRjXlfCiI,378
 langroid/language_models/mock_lm.py,sha256=5BgHKDVRWFbUwDT_PFgTZXz9-k8wJSA2e3PZmyDgQ1k,4022
-langroid/language_models/openai_gpt.py,sha256=gkF3sQsBrS8_x4x0l6GLdF_5MlNRdsY5hAmRPSsKAKE,75320
+langroid/language_models/openai_gpt.py,sha256=MW-eOc8bFVAvnJZaj6iTi8NYJOR0hcAx7ZbBuY4mFfY,76410
 langroid/language_models/prompt_formatter/__init__.py,sha256=2-5cdE24XoFDhifOLl8yiscohil1ogbP1ECkYdBlBsk,372
 langroid/language_models/prompt_formatter/base.py,sha256=eDS1sgRNZVnoajwV_ZIha6cba5Dt8xjgzdRbPITwx3Q,1221
 langroid/language_models/prompt_formatter/hf_formatter.py,sha256=PVJppmjRvD-2DF-XNC6mE05vTZ9wbu37SmXwZBQhad0,5055
 langroid/language_models/prompt_formatter/llama2_formatter.py,sha256=YdcO88qyBeuMENVIVvVqSYuEpvYSTndUe_jd6hVTko4,2899
-langroid/language_models/utils.py,sha256=o6Zo2cnnvKrfSgF26knVQ1xkSxEoE7yN85296gNdVOw,4858
+langroid/language_models/utils.py,sha256=L4_CbihDMTGcsg0TOG1Yd5JFEto46--h7CX_14m89sQ,5016
 langroid/mytypes.py,sha256=ptAFxEAtiwmIfUnGisNotTe8wT9LKBf22lOfPgZoQIY,2368
 langroid/parsing/__init__.py,sha256=ZgSAfgTC6VsTLFlRSWT-TwYco7SQeRMeZG-49MnKYGY,936
 langroid/parsing/agent_chats.py,sha256=sbZRV9ujdM5QXvvuHVjIi2ysYSYlap-uqfMMUKulrW0,1068
@@ -155,8 +155,8 @@ langroid/vector_store/meilisearch.py,sha256=6frB7GFWeWmeKzRfLZIvzRjllniZ1cYj3Hmh
 langroid/vector_store/momento.py,sha256=UNHGT6jXuQtqY9f6MdqGU14bVnS0zHgIJUa30ULpUJo,10474
 langroid/vector_store/qdrant_cloud.py,sha256=3im4Mip0QXLkR6wiqVsjV1QvhSElfxdFSuDKddBDQ-4,188
 langroid/vector_store/qdrantdb.py,sha256=v7mCsijc2GdRJyil-yFaUVAX4SX5D75mD3vzlpjCMuo,17393
-pyproject.toml,sha256=JrFkZ_N6bpaZki0eTQcBGhaY6WS0l3fc-fvsepT1WtA,7525
-langroid-0.31.3.dist-info/LICENSE,sha256=EgVbvA6VSYgUlvC3RvPKehSg7MFaxWDsFuzLOsPPfJg,1065
-langroid-0.31.3.dist-info/METADATA,sha256=G-Y0UMCtjvcNeV8l3aHBQvXGuurDMe8qNC-rX5geazg,58250
-langroid-0.31.3.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-langroid-0.31.3.dist-info/RECORD,,
+pyproject.toml,sha256=7nOFChBTyZrJaBWcdJ4k9_99dVpYzP3XMKoU3UWrtsw,7525
+langroid-0.32.0.dist-info/LICENSE,sha256=EgVbvA6VSYgUlvC3RvPKehSg7MFaxWDsFuzLOsPPfJg,1065
+langroid-0.32.0.dist-info/METADATA,sha256=BVnnbSEMCLkiirANizar45wRF-cECcvZ7tz-yy2kczk,58250
+langroid-0.32.0.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+langroid-0.32.0.dist-info/RECORD,,

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "langroid"
-version = "0.31.3"
+version = "0.32.0"
 description = "Harness LLMs with Multi-Agent Programming"
 authors = ["Prasad Chalasani <pchalasani@gmail.com>"]
 readme = "README.md"