langroid 0.28.7__py3-none-any.whl → 0.30.0__py3-none-any.whl

langroid/agent/.chainlit/translations/zh-CN.json

@@ -0,0 +1,229 @@
+{
+    "components": {
+        "atoms": {
+            "buttons": {
+                "userButton": {
+                    "menu": {
+                        "settings": "\u8bbe\u7f6e",
+                        "settingsKey": "S",
+                        "APIKeys": "API \u5bc6\u94a5",
+                        "logout": "\u767b\u51fa"
+                    }
+                }
+            }
+        },
+        "molecules": {
+            "newChatButton": {
+                "newChat": "\u65b0\u5efa\u5bf9\u8bdd"
+            },
+            "tasklist": {
+                "TaskList": {
+                    "title": "\ud83d\uddd2\ufe0f \u4efb\u52a1\u5217\u8868",
+                    "loading": "\u52a0\u8f7d\u4e2d...",
+                    "error": "\u53d1\u751f\u9519\u8bef"
+                }
+            },
+            "attachments": {
+                "cancelUpload": "\u53d6\u6d88\u4e0a\u4f20",
+                "removeAttachment": "\u79fb\u9664\u9644\u4ef6"
+            },
+            "newChatDialog": {
+                "createNewChat": "\u521b\u5efa\u65b0\u5bf9\u8bdd\uff1f",
+                "clearChat": "\u8fd9\u5c06\u6e05\u9664\u5f53\u524d\u6d88\u606f\u5e76\u5f00\u59cb\u65b0\u7684\u5bf9\u8bdd\u3002",
+                "cancel": "\u53d6\u6d88",
+                "confirm": "\u786e\u8ba4"
+            },
+            "settingsModal": {
+                "settings": "\u8bbe\u7f6e",
+                "expandMessages": "\u5c55\u5f00\u6d88\u606f",
+                "hideChainOfThought": "\u9690\u85cf\u601d\u8003\u94fe",
+                "darkMode": "\u6697\u8272\u6a21\u5f0f"
+            },
+            "detailsButton": {
+                "using": "\u4f7f\u7528",
+                "used": "\u5df2\u7528"
+            },
+            "auth": {
+                "authLogin": {
+                    "title": "\u767b\u5f55\u4ee5\u8bbf\u95ee\u5e94\u7528\u3002",
+                    "form": {
+                        "email": "\u7535\u5b50\u90ae\u7bb1\u5730\u5740",
+                        "password": "\u5bc6\u7801",
+                        "noAccount": "\u6ca1\u6709\u8d26\u6237\uff1f",
+                        "alreadyHaveAccount": "\u5df2\u6709\u8d26\u6237\uff1f",
+                        "signup": "\u6ce8\u518c",
+                        "signin": "\u767b\u5f55",
+                        "or": "\u6216\u8005",
+                        "continue": "\u7ee7\u7eed",
+                        "forgotPassword": "\u5fd8\u8bb0\u5bc6\u7801\uff1f",
+                        "passwordMustContain": "\u60a8\u7684\u5bc6\u7801\u5fc5\u987b\u5305\u542b\uff1a",
+                        "emailRequired": "\u7535\u5b50\u90ae\u7bb1\u662f\u5fc5\u586b\u9879",
+                        "passwordRequired": "\u5bc6\u7801\u662f\u5fc5\u586b\u9879"
+                    },
+                    "error": {
+                        "default": "\u65e0\u6cd5\u767b\u5f55\u3002",
+                        "signin": "\u5c1d\u8bd5\u4f7f\u7528\u4e0d\u540c\u7684\u8d26\u6237\u767b\u5f55\u3002",
+                        "oauthsignin": "\u5c1d\u8bd5\u4f7f\u7528\u4e0d\u540c\u7684\u8d26\u6237\u767b\u5f55\u3002",
+                        "redirect_uri_mismatch": "\u91cd\u5b9a\u5411URI\u4e0eOAuth\u5e94\u7528\u914d\u7f6e\u4e0d\u5339\u914d\u3002",
+                        "oauthcallbackerror": "\u5c1d\u8bd5\u4f7f\u7528\u4e0d\u540c\u7684\u8d26\u6237\u767b\u5f55\u3002",
+                        "oauthcreateaccount": "\u5c1d\u8bd5\u4f7f\u7528\u4e0d\u540c\u7684\u8d26\u6237\u767b\u5f55\u3002",
+                        "emailcreateaccount": "\u5c1d\u8bd5\u4f7f\u7528\u4e0d\u540c\u7684\u8d26\u6237\u767b\u5f55\u3002",
+                        "callback": "\u5c1d\u8bd5\u4f7f\u7528\u4e0d\u540c\u7684\u8d26\u6237\u767b\u5f55\u3002",
+                        "oauthaccountnotlinked": "\u4e3a\u4e86\u9a8c\u8bc1\u60a8\u7684\u8eab\u4efd\uff0c\u8bf7\u4f7f\u7528\u6700\u521d\u4f7f\u7528\u7684\u540c\u4e00\u8d26\u6237\u767b\u5f55\u3002",
+                        "emailsignin": "\u65e0\u6cd5\u53d1\u9001\u90ae\u4ef6\u3002",
+                        "emailverify": "\u8bf7\u9a8c\u8bc1\u60a8\u7684\u7535\u5b50\u90ae\u4ef6\uff0c\u5df2\u53d1\u9001\u4e00\u5c01\u65b0\u90ae\u4ef6\u3002",
+                        "credentialssignin": "\u767b\u5f55\u5931\u8d25\u3002\u8bf7\u68c0\u67e5\u60a8\u63d0\u4f9b\u7684\u8be6\u7ec6\u4fe1\u606f\u662f\u5426\u6b63\u786e\u3002",
+                        "sessionrequired": "\u8bf7\u767b\u5f55\u4ee5\u8bbf\u95ee\u6b64\u9875\u9762\u3002"
+                    }
+                },
+                "authVerifyEmail": {
+                    "almostThere": "\u60a8\u5feb\u6210\u529f\u4e86\uff01\u6211\u4eec\u5df2\u5411 ",
+                    "verifyEmailLink": "\u8bf7\u5355\u51fb\u8be5\u90ae\u4ef6\u4e2d\u7684\u94fe\u63a5\u4ee5\u5b8c\u6210\u6ce8\u518c\u3002",
+                    "didNotReceive": "\u6ca1\u627e\u5230\u90ae\u4ef6\uff1f",
+                    "resendEmail": "\u91cd\u65b0\u53d1\u9001\u90ae\u4ef6",
+                    "goBack": "\u8fd4\u56de",
+                    "emailSent": "\u90ae\u4ef6\u5df2\u6210\u529f\u53d1\u9001\u3002",
+                    "verifyEmail": "\u9a8c\u8bc1\u60a8\u7684\u7535\u5b50\u90ae\u4ef6\u5730\u5740"
+                },
+                "providerButton": {
+                    "continue": "\u4f7f\u7528{{provider}}\u7ee7\u7eed",
+                    "signup": "\u4f7f\u7528{{provider}}\u6ce8\u518c"
+                },
+                "authResetPassword": {
+                    "newPasswordRequired": "\u65b0\u5bc6\u7801\u662f\u5fc5\u586b\u9879",
+                    "passwordsMustMatch": "\u5bc6\u7801\u5fc5\u987b\u4e00\u81f4",
+                    "confirmPasswordRequired": "\u786e\u8ba4\u5bc6\u7801\u662f\u5fc5\u586b\u9879",
+                    "newPassword": "\u65b0\u5bc6\u7801",
+                    "confirmPassword": "\u786e\u8ba4\u5bc6\u7801",
+                    "resetPassword": "\u91cd\u7f6e\u5bc6\u7801"
+                },
+                "authForgotPassword": {
+                    "email": "\u7535\u5b50\u90ae\u7bb1\u5730\u5740",
+                    "emailRequired": "\u7535\u5b50\u90ae\u7bb1\u662f\u5fc5\u586b\u9879",
+                    "emailSent": "\u8bf7\u68c0\u67e5\u7535\u5b50\u90ae\u7bb1{{email}}\u4ee5\u83b7\u53d6\u91cd\u7f6e\u5bc6\u7801\u7684\u6307\u793a\u3002",
+                    "enterEmail": "\u8bf7\u8f93\u5165\u60a8\u7684\u7535\u5b50\u90ae\u7bb1\u5730\u5740\uff0c\u6211\u4eec\u5c06\u53d1\u9001\u91cd\u7f6e\u5bc6\u7801\u7684\u6307\u793a\u3002",
+                    "resendEmail": "\u91cd\u65b0\u53d1\u9001\u90ae\u4ef6",
+                    "continue": "\u7ee7\u7eed",
+                    "goBack": "\u8fd4\u56de"
+                }
+            }
+        },
+        "organisms": {
+            "chat": {
+                "history": {
+                    "index": {
+                        "showHistory": "\u663e\u793a\u5386\u53f2",
+                        "lastInputs": "\u6700\u540e\u8f93\u5165",
+                        "noInputs": "\u5982\u6b64\u7a7a\u65f7...",
+                        "loading": "\u52a0\u8f7d\u4e2d..."
+                    }
+                },
+                "inputBox": {
+                    "input": {
+                        "placeholder": "\u5728\u8fd9\u91cc\u8f93\u5165\u60a8\u7684\u6d88\u606f..."
+                    },
+                    "speechButton": {
+                        "start": "\u5f00\u59cb\u5f55\u97f3",
+                        "stop": "\u505c\u6b62\u5f55\u97f3"
+                    },
+                    "SubmitButton": {
+                        "sendMessage": "\u53d1\u9001\u6d88\u606f",
+                        "stopTask": "\u505c\u6b62\u4efb\u52a1"
+                    },
+                    "UploadButton": {
+                        "attachFiles": "\u9644\u52a0\u6587\u4ef6"
+                    },
+                    "waterMark": {
+                        "text": "\u4f7f\u7528"
+                    }
+                },
+                "Messages": {
+                    "index": {
+                        "running": "\u8fd0\u884c\u4e2d",
+                        "executedSuccessfully": "\u6267\u884c\u6210\u529f",
+                        "failed": "\u5931\u8d25",
+                        "feedbackUpdated": "\u53cd\u9988\u66f4\u65b0",
+                        "updating": "\u6b63\u5728\u66f4\u65b0"
+                    }
+                },
+                "dropScreen": {
+                    "dropYourFilesHere": "\u5728\u8fd9\u91cc\u62d6\u653e\u60a8\u7684\u6587\u4ef6"
+                },
+                "index": {
+                    "failedToUpload": "\u4e0a\u4f20\u5931\u8d25",
+                    "cancelledUploadOf": "\u53d6\u6d88\u4e0a\u4f20",
+                    "couldNotReachServer": "\u65e0\u6cd5\u8fde\u63a5\u5230\u670d\u52a1\u5668",
+                    "continuingChat": "\u7ee7\u7eed\u4e4b\u524d\u7684\u5bf9\u8bdd"
+                },
+                "settings": {
+                    "settingsPanel": "\u8bbe\u7f6e\u9762\u677f",
+                    "reset": "\u91cd\u7f6e",
+                    "cancel": "\u53d6\u6d88",
+                    "confirm": "\u786e\u8ba4"
+                }
+            },
+            "threadHistory": {
+                "sidebar": {
+                    "filters": {
+                        "FeedbackSelect": {
+                            "feedbackAll": "\u53cd\u9988\uff1a\u5168\u90e8",
+                            "feedbackPositive": "\u53cd\u9988\uff1a\u6b63\u9762",
+                            "feedbackNegative": "\u53cd\u9988\uff1a\u8d1f\u9762"
+                        },
+                        "SearchBar": {
+                            "search": "\u641c\u7d22"
+                        }
+                    },
+                    "DeleteThreadButton": {
+                        "confirmMessage": "\u8fd9\u5c06\u5220\u9664\u7ebf\u7a0b\u53ca\u5176\u6d88\u606f\u548c\u5143\u7d20\u3002",
+                        "cancel": "\u53d6\u6d88",
+                        "confirm": "\u786e\u8ba4",
+                        "deletingChat": "\u5220\u9664\u5bf9\u8bdd",
+                        "chatDeleted": "\u5bf9\u8bdd\u5df2\u5220\u9664"
+                    },
+                    "index": {
+                        "pastChats": "\u8fc7\u5f80\u5bf9\u8bdd"
+                    },
+                    "ThreadList": {
+                        "empty": "\u7a7a\u7684...",
+                        "today": "\u4eca\u5929",
+                        "yesterday": "\u6628\u5929",
+                        "previous7days": "\u524d7\u5929",
+                        "previous30days": "\u524d30\u5929"
+                    },
+                    "TriggerButton": {
+                        "closeSidebar": "\u5173\u95ed\u4fa7\u8fb9\u680f",
+                        "openSidebar": "\u6253\u5f00\u4fa7\u8fb9\u680f"
+                    }
+                },
+                "Thread": {
+                    "backToChat": "\u8fd4\u56de\u5bf9\u8bdd",
+                    "chatCreatedOn": "\u6b64\u5bf9\u8bdd\u521b\u5efa\u4e8e"
+                }
+            },
+            "header": {
+                "chat": "\u5bf9\u8bdd",
+                "readme": "\u8bf4\u660e"
+            }
+        }
+    },
+    "hooks": {
+        "useLLMProviders": {
+            "failedToFetchProviders": "\u83b7\u53d6\u63d0\u4f9b\u8005\u5931\u8d25:"
+        }
+    },
+    "pages": {
+        "Design": {},
+        "Env": {
+            "savedSuccessfully": "\u4fdd\u5b58\u6210\u529f",
+            "requiredApiKeys": "\u5fc5\u9700\u7684API\u5bc6\u94a5",
+            "requiredApiKeysInfo": "\u8981\u4f7f\u7528\u6b64\u5e94\u7528\uff0c\u9700\u8981\u4ee5\u4e0bAPI\u5bc6\u94a5\u3002\u8fd9\u4e9b\u5bc6\u94a5\u5b58\u50a8\u5728\u60a8\u7684\u8bbe\u5907\u672c\u5730\u5b58\u50a8\u4e2d\u3002"
+        },
+        "Page": {
+            "notPartOfProject": "\u60a8\u4e0d\u662f\u6b64\u9879\u76ee\u7684\u4e00\u90e8\u5206\u3002"
+        },
+        "ResumeButton": {
+            "resumeChat": "\u6062\u590d\u5bf9\u8bdd"
+        }
+    }
+}

langroid/embedding_models/__init__.py

@@ -9,8 +9,10 @@ from .base import (
 from .models import (
     OpenAIEmbeddings,
     OpenAIEmbeddingsConfig,
-    SentenceTransformerEmbeddingsConfig,
     SentenceTransformerEmbeddings,
+    SentenceTransformerEmbeddingsConfig,
+    LlamaCppServerEmbeddings,
+    LlamaCppServerEmbeddingsConfig,
     embedding_model,
 )
 from .remote_embeds import (
@@ -27,8 +29,10 @@ __all__ = [
     "EmbeddingModelsConfig",
     "OpenAIEmbeddings",
     "OpenAIEmbeddingsConfig",
-    "SentenceTransformerEmbeddingsConfig",
     "SentenceTransformerEmbeddings",
+    "SentenceTransformerEmbeddingsConfig",
+    "LlamaCppServerEmbeddings",
+    "LlamaCppServerEmbeddingsConfig",
     "embedding_model",
     "RemoteEmbeddingsConfig",
     "RemoteEmbeddings",

langroid/embedding_models/base.py

@@ -26,6 +26,8 @@ class EmbeddingModel(ABC):
         from langroid.embedding_models.models import (
             FastEmbedEmbeddings,
             FastEmbedEmbeddingsConfig,
+            LlamaCppServerEmbeddings,
+            LlamaCppServerEmbeddingsConfig,
             OpenAIEmbeddings,
             OpenAIEmbeddingsConfig,
             SentenceTransformerEmbeddings,
@@ -44,6 +46,8 @@ class EmbeddingModel(ABC):
             return SentenceTransformerEmbeddings(config)
         elif isinstance(config, FastEmbedEmbeddingsConfig):
             return FastEmbedEmbeddings(config)
+        elif isinstance(config, LlamaCppServerEmbeddingsConfig):
+            return LlamaCppServerEmbeddings(config)
         else:
             raise ValueError(f"Unknown embedding config: {config.__repr_name__}")
 

langroid/embedding_models/models.py

@@ -3,6 +3,7 @@ import os
 from functools import cached_property
 from typing import Any, Callable, Dict, List, Optional
 
+import requests
 import tiktoken
 from dotenv import load_dotenv
 from openai import OpenAI
@@ -48,13 +49,19 @@ class FastEmbedEmbeddingsConfig(EmbeddingModelsConfig):
     additional_kwargs: Dict[str, Any] = {}
 
 
+class LlamaCppServerEmbeddingsConfig(EmbeddingModelsConfig):
+    api_base: str = ""
+    context_length: int = 2048
+    batch_size: int = 2048
+
+
 class EmbeddingFunctionCallable:
     """
     A callable class designed to generate embeddings for a list of texts using
     the OpenAI API, with automatic retries on failure.
 
     Attributes:
-        model (OpenAIEmbeddings): An instance of OpenAIEmbeddings that provides
+        embed_model (EmbeddingModel): An instance of EmbeddingModel that provides
                                 configuration and utilities for generating embeddings.
 
     Methods:
@@ -62,7 +69,7 @@ class EmbeddingFunctionCallable:
                                 a list of input texts.
     """
 
-    def __init__(self, model: "OpenAIEmbeddings", batch_size: int = 512):
+    def __init__(self, embed_model: EmbeddingModel, batch_size: int = 512):
         """
         Initialize the EmbeddingFunctionCallable with a specific model.
 
@@ -71,7 +78,7 @@ class EmbeddingFunctionCallable:
             generating embeddings.
             batch_size (int): Batch size
         """
-        self.model = model
+        self.embed_model = embed_model
         self.batch_size = batch_size
 
     def __call__(self, input: List[str]) -> Embeddings:
@@ -91,14 +98,46 @@ class EmbeddingFunctionCallable:
         Returns:
             Embeddings: A list of embedding vectors corresponding to the input texts.
         """
-        tokenized_texts = self.model.truncate_texts(input)
         embeds = []
-        for batch in batched(tokenized_texts, self.batch_size):
-            result = self.model.client.embeddings.create(
-                input=batch, model=self.model.config.model_name
+
+        if isinstance(self.embed_model, OpenAIEmbeddings):
+            tokenized_texts = self.embed_model.truncate_texts(input)
+
+            for batch in batched(tokenized_texts, self.batch_size):
+                result = self.embed_model.client.embeddings.create(
+                    input=batch, model=self.embed_model.config.model_name
+                )
+                batch_embeds = [d.embedding for d in result.data]
+                embeds.extend(batch_embeds)
+
+        elif isinstance(self.embed_model, SentenceTransformerEmbeddings):
+            if self.embed_model.config.data_parallel:
+                embeds = self.embed_model.model.encode_multi_process(
+                    input,
+                    self.embed_model.pool,
+                    batch_size=self.batch_size,
+                ).tolist()
+            else:
+                for str_batch in batched(input, self.batch_size):
+                    batch_embeds = self.embed_model.model.encode(
+                        str_batch, convert_to_numpy=True
+                    ).tolist()  # type: ignore
+                    embeds.extend(batch_embeds)
+
+        elif isinstance(self.embed_model, FastEmbedEmbeddings):
+            embeddings = self.embed_model.model.embed(
+                input, batch_size=self.batch_size, parallel=self.embed_model.parallel
             )
-            batch_embeds = [d.embedding for d in result.data]
-            embeds.extend(batch_embeds)
+
+            embeds = [embedding.tolist() for embedding in embeddings]
+        elif isinstance(self.embed_model, LlamaCppServerEmbeddings):
+            for input_string in input:
+                tokenized_text = self.embed_model.tokenize_string(input_string)
+                for token_batch in batched(tokenized_text, self.batch_size):
+                    gen_embedding = self.embed_model.generate_embedding(
+                        self.embed_model.detokenize_string(list(token_batch))
+                    )
+                    embeds.append(gen_embedding)
         return embeds
 
 
@@ -176,24 +215,7 @@ class SentenceTransformerEmbeddings(EmbeddingModel):
         self.config.context_length = self.tokenizer.model_max_length
 
     def embedding_fn(self) -> Callable[[List[str]], Embeddings]:
-        def fn(texts: List[str]) -> Embeddings:
-            if self.config.data_parallel:
-                embeds: Embeddings = self.model.encode_multi_process(
-                    texts,
-                    self.pool,
-                    batch_size=self.config.batch_size,
-                ).tolist()
-            else:
-                embeds = []
-                for batch in batched(texts, self.config.batch_size):
-                    batch_embeds = self.model.encode(
-                        batch, convert_to_numpy=True
-                    ).tolist()  # type: ignore
-                    embeds.extend(batch_embeds)
-
-            return embeds
-
-        return fn
+        return EmbeddingFunctionCallable(self, self.config.batch_size)
 
     @property
     def embedding_dims(self) -> int:
@@ -214,10 +236,10 @@ class FastEmbedEmbeddings(EmbeddingModel):
 
         super().__init__()
         self.config = config
-        self._batch_size = config.batch_size
-        self._parallel = config.parallel
+        self.batch_size = config.batch_size
+        self.parallel = config.parallel
 
-        self._model = TextEmbedding(
+        self.model = TextEmbedding(
             model_name=self.config.model_name,
             cache_dir=self.config.cache_dir,
             threads=self.config.threads,
@@ -225,14 +247,7 @@ class FastEmbedEmbeddings(EmbeddingModel):
         )
 
     def embedding_fn(self) -> Callable[[List[str]], Embeddings]:
-        def fn(texts: List[str]) -> Embeddings:
-            embeddings = self._model.embed(
-                texts, batch_size=self._batch_size, parallel=self._parallel
-            )
-
-            return [embedding.tolist() for embedding in embeddings]
-
-        return fn
+        return EmbeddingFunctionCallable(self, self.config.batch_size)
 
     @cached_property
     def embedding_dims(self) -> int:
@@ -240,10 +255,105 @@ class FastEmbedEmbeddings(EmbeddingModel):
         return len(embed_func(["text"])[0])
 
 
+LCSEC = LlamaCppServerEmbeddingsConfig
+
+
+class LlamaCppServerEmbeddings(EmbeddingModel):
+    def __init__(self, config: LCSEC = LCSEC()):
+        super().__init__()
+        self.config = config
+
+        if self.config.api_base == "":
+            raise ValueError(
+                """Api Base MUST be set for Llama Server Embeddings.
+                """
+            )
+
+        self.tokenize_url = self.config.api_base + "/tokenize"
+        self.detokenize_url = self.config.api_base + "/detokenize"
+        self.embedding_url = self.config.api_base + "/embeddings"
+
+    def tokenize_string(self, text: str) -> List[int]:
+        data = {"content": text, "add_special": False, "with_pieces": False}
+        response = requests.post(self.tokenize_url, json=data)
+
+        if response.status_code == 200:
+            tokens = response.json()["tokens"]
+            if not (isinstance(tokens, list) and isinstance(tokens[0], (int, float))):
+                # not all(isinstance(token, (int, float)) for token in tokens):
+                raise ValueError(
+                    """Tokenizer endpoint has not returned the correct format. 
+                   Is the URL correct?
+                """
+                )
+            return tokens
+        else:
+            raise requests.HTTPError(
+                self.tokenize_url,
+                response.status_code,
+                "Failed to connect to tokenization provider",
+            )
+
+    def detokenize_string(self, tokens: List[int]) -> str:
+        data = {"tokens": tokens}
+        response = requests.post(self.detokenize_url, json=data)
+
+        if response.status_code == 200:
+            text = response.json()["content"]
+            if not isinstance(text, str):
+                raise ValueError(
+                    """Deokenizer endpoint has not returned the correct format. 
+                   Is the URL correct?
+                """
+                )
+            return text
+        else:
+            raise requests.HTTPError(
+                self.detokenize_url,
+                response.status_code,
+                "Failed to connect to detokenization provider",
+            )
+
+    def truncate_string_to_context_size(self, text: str) -> str:
+        tokens = self.tokenize_string(text)
+        tokens = tokens[: self.config.context_length]
+        return self.detokenize_string(tokens)
+
+    def generate_embedding(self, text: str) -> List[int | float]:
+        data = {"content": text}
+        response = requests.post(self.embedding_url, json=data)
+
+        if response.status_code == 200:
+            embeddings = response.json()["embedding"]
+            if not (
+                isinstance(embeddings, list) and isinstance(embeddings[0], (int, float))
+            ):
+                raise ValueError(
+                    """Embedding endpoint has not returned the correct format. 
+                   Is the URL correct?
+                """
+                )
+            return embeddings
+        else:
+            raise requests.HTTPError(
+                self.embedding_url,
+                response.status_code,
+                "Failed to connect to embedding provider",
+            )
+
+    def embedding_fn(self) -> Callable[[List[str]], Embeddings]:
+        return EmbeddingFunctionCallable(self, self.config.batch_size)
+
+    @property
+    def embedding_dims(self) -> int:
+        return self.config.dims
+
+
 def embedding_model(embedding_fn_type: str = "openai") -> EmbeddingModel:
     """
     Args:
-        embedding_fn_type: "openai" or "sentencetransformer" # others soon
+        embedding_fn_type: "openai" or "fastembed" or
+                           "llamacppserver" or "sentencetransformer" # others soon
     Returns:
         EmbeddingModel
     """
@@ -251,5 +361,7 @@ def embedding_model(embedding_fn_type: str = "openai") -> EmbeddingModel:
         return OpenAIEmbeddings  # type: ignore
     elif embedding_fn_type == "fastembed":
         return FastEmbedEmbeddings  # type: ignore
+    elif embedding_fn_type == "llamacppserver":
+        return LlamaCppServerEmbeddings  # type: ignore
     else:  # default sentence transformer
         return SentenceTransformerEmbeddings  # type: ignore

langroid/language_models/azure_openai.py

@@ -1,3 +1,6 @@
+import logging
+from typing import Callable
+
 from dotenv import load_dotenv
 from httpx import Timeout
 from openai import AsyncAzureOpenAI, AzureOpenAI
@@ -15,6 +18,8 @@ azureStructuredOutputList = [
 
 azureStructuredOutputAPIMin = "2024-08-01-preview"
 
+logger = logging.getLogger(__name__)
+
 
 class AzureConfig(OpenAIGPTConfig):
     """
@@ -42,6 +47,10 @@ class AzureConfig(OpenAIGPTConfig):
     model_version: str = ""  # is used to determine the cost of using the model
     api_base: str = ""
 
+    # Alternatively, bring your own clients:
+    azure_openai_client_provider: Callable[[], AzureOpenAI] | None = None
+    azure_openai_async_client_provider: Callable[[], AsyncAzureOpenAI] | None = None
+
     # all of the vars above can be set via env vars,
     # by upper-casing the name and prefixing with `env_prefix`, e.g.
     # AZURE_OPENAI_API_VERSION=2023-05-15
@@ -69,20 +78,6 @@ class AzureGPT(OpenAIGPT):
         load_dotenv()
         super().__init__(config)
         self.config: AzureConfig = config
-        if self.config.api_key == "":
-            raise ValueError(
-                """
-                AZURE_OPENAI_API_KEY not set in .env file,
-                please set it to your Azure API key."""
-            )
-
-        if self.config.api_base == "":
-            raise ValueError(
-                """
-                AZURE_OPENAI_API_BASE not set in .env file,
-                please set it to your Azure API key."""
-            )
-
         if self.config.deployment_name == "":
             raise ValueError(
                 """
@@ -98,6 +93,57 @@ class AzureGPT(OpenAIGPT):
                 please set it to chat model name in your deployment."""
             )
 
+        if (
+            self.config.azure_openai_client_provider
+            or self.config.azure_openai_async_client_provider
+        ):
+            if not self.config.azure_openai_client_provider:
+                self.client = None
+                logger.warning(
+                    "Using user-provided Azure OpenAI client, but only async "
+                    "client has been provided. Synchronous calls will fail."
+                )
+            if not self.config.azure_openai_async_client_provider:
+                self.async_client = None
+                logger.warning(
+                    "Using user-provided Azure OpenAI client, but no async "
+                    "client has been provided. Asynchronous calls will fail."
+                )
+
+            if self.config.azure_openai_client_provider:
+                self.client = self.config.azure_openai_client_provider()
+            if self.config.azure_openai_async_client_provider:
+                self.async_client = self.config.azure_openai_async_client_provider()
+                self.async_client.timeout = Timeout(self.config.timeout)
+        else:
+            if self.config.api_key == "":
+                raise ValueError(
+                    """
+                    AZURE_OPENAI_API_KEY not set in .env file,
+                    please set it to your Azure API key."""
+                )
+
+            if self.config.api_base == "":
+                raise ValueError(
+                    """
+                    AZURE_OPENAI_API_BASE not set in .env file,
+                    please set it to your Azure API key."""
+                )
+
+            self.client = AzureOpenAI(
+                api_key=self.config.api_key,
+                azure_endpoint=self.config.api_base,
+                api_version=self.config.api_version,
+                azure_deployment=self.config.deployment_name,
+            )
+            self.async_client = AsyncAzureOpenAI(
+                api_key=self.config.api_key,
+                azure_endpoint=self.config.api_base,
+                api_version=self.config.api_version,
+                azure_deployment=self.config.deployment_name,
+                timeout=Timeout(self.config.timeout),
+            )
+
         # set the chat model to be the same as the model_name
         # This corresponds to the gpt model you chose for your deployment
         # when you deployed a model
@@ -108,20 +154,6 @@ class AzureGPT(OpenAIGPT):
             and self.config.model_version in azureStructuredOutputList
         )
 
-        self.client = AzureOpenAI(
-            api_key=self.config.api_key,
-            azure_endpoint=self.config.api_base,
-            api_version=self.config.api_version,
-            azure_deployment=self.config.deployment_name,
-        )
-        self.async_client = AsyncAzureOpenAI(
-            api_key=self.config.api_key,
-            azure_endpoint=self.config.api_base,
-            api_version=self.config.api_version,
-            azure_deployment=self.config.deployment_name,
-            timeout=Timeout(self.config.timeout),
-        )
-
     def set_chat_model(self) -> None:
         """
         Sets the chat model configuration based on the model name specified in the