langroid 0.2.11__tar.gz → 0.3.0__tar.gz

{langroid-0.2.11 → langroid-0.3.0}/PKG-INFO

@@ -1,11 +1,11 @@
 Metadata-Version: 2.1
 Name: langroid
-Version: 0.2.11
+Version: 0.3.0
 Summary: Harness LLMs with Multi-Agent Programming
 License: MIT
 Author: Prasad Chalasani
 Author-email: pchalasani@gmail.com
-Requires-Python: >=3.10,<4.0
+Requires-Python: >=3.10,<3.13
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
@@ -17,6 +17,7 @@ Provides-Extra: chromadb
 Provides-Extra: db
 Provides-Extra: doc-chat
 Provides-Extra: docx
+Provides-Extra: fastembed
 Provides-Extra: hf-embeddings
 Provides-Extra: hf-transformers
 Provides-Extra: lancedb
@@ -44,6 +45,7 @@ Requires-Dist: docstring-parser (>=0.15,<0.16)
 Requires-Dist: duckduckgo-search (>=6.0.0,<7.0.0)
 Requires-Dist: faker (>=18.9.0,<19.0.0)
 Requires-Dist: fakeredis (>=2.12.1,<3.0.0)
+Requires-Dist: fastembed (>=0.3.1,<0.4.0) ; extra == "all" or extra == "fastembed"
 Requires-Dist: fire (>=0.5.0,<0.6.0)
 Requires-Dist: google-api-python-client (>=2.95.0,<3.0.0)
 Requires-Dist: google-generativeai (>=0.5.2,<0.6.0)

{langroid-0.2.11 → langroid-0.3.0}/langroid/agent/special/sql/sql_chat_agent.py

@@ -66,6 +66,17 @@ Start by asking what I would like to know about the data.
 
 """
 
+ADDRESSING_INSTRUCTION = """
+IMPORTANT - Whenever you are NOT writing a SQL query, make sure you address the user
+using {prefix}User. You MUST use the EXACT syntax {prefix} !!!
+
+In other words, you ALWAYS write EITHER:
+ - a SQL query using the `run_query` tool, 
+ - OR address the user using {prefix}User.
+
+"""
+
+
 SQL_ERROR_MSG = "There was an error in your SQL Query"
 
 
@@ -81,6 +92,7 @@ class SQLChatAgentConfig(ChatAgentConfig):
     context_descriptions: Dict[str, Dict[str, Union[str, Dict[str, str]]]] = {}
     use_schema_tools: bool = False
     multi_schema: bool = False
+    addressing_prefix: str = ""
 
     """
     Optional, but strongly recommended, context descriptions for tables, columns, 
@@ -206,6 +218,10 @@ class SQLChatAgent(ChatAgent):
         """Initialize message tools used for chatting."""
         message = self._format_message()
         self.config.system_message = self.config.system_message.format(mode=message)
+        if self.config.addressing_prefix != "":
+            self.config.system_message += ADDRESSING_INSTRUCTION.format(
+                prefix=self.config.addressing_prefix
+            )
         super().__init__(self.config)
         self.enable_message(RunQueryTool)
         if self.config.use_schema_tools:

{langroid-0.2.11 → langroid-0.3.0}/langroid/embedding_models/base.py

@@ -24,6 +24,8 @@ class EmbeddingModel(ABC):
     @classmethod
     def create(cls, config: EmbeddingModelsConfig) -> "EmbeddingModel":
         from langroid.embedding_models.models import (
+            FastEmbedEmbeddings,
+            FastEmbedEmbeddingsConfig,
             OpenAIEmbeddings,
             OpenAIEmbeddingsConfig,
             SentenceTransformerEmbeddings,
@@ -40,6 +42,8 @@ class EmbeddingModel(ABC):
             return OpenAIEmbeddings(config)
         elif isinstance(config, SentenceTransformerEmbeddingsConfig):
             return SentenceTransformerEmbeddings(config)
+        elif isinstance(config, FastEmbedEmbeddingsConfig):
+            return FastEmbedEmbeddings(config)
         else:
             raise ValueError(f"Unknown embedding config: {config.__repr_name__}")
 

{langroid-0.2.11 → langroid-0.3.0}/langroid/embedding_models/models.py

@@ -1,12 +1,14 @@
 import atexit
 import os
-from typing import Callable, List, Optional
+from functools import cached_property
+from typing import Any, Callable, Dict, List, Optional
 
 import tiktoken
 from dotenv import load_dotenv
 from openai import OpenAI
 
 from langroid.embedding_models.base import EmbeddingModel, EmbeddingModelsConfig
+from langroid.exceptions import LangroidImportError
 from langroid.mytypes import Embeddings
 from langroid.parsing.utils import batched
 
@@ -32,6 +34,20 @@ class SentenceTransformerEmbeddingsConfig(EmbeddingModelsConfig):
     devices: Optional[list[str]] = None
 
 
+class FastEmbedEmbeddingsConfig(EmbeddingModelsConfig):
+    """Config for qdrant/fastembed embeddings,
+    see here: https://github.com/qdrant/fastembed
+    """
+
+    model_type: str = "fastembed"
+    model_name: str = "BAAI/bge-small-en-v1.5"
+    batch_size: int = 256
+    cache_dir: Optional[str] = None
+    threads: Optional[int] = None
+    parallel: Optional[int] = None
+    additional_kwargs: Dict[str, Any] = {}
+
+
 class EmbeddingFunctionCallable:
     """
     A callable class designed to generate embeddings for a list of texts using
@@ -189,6 +205,41 @@ class SentenceTransformerEmbeddings(EmbeddingModel):
         return dims  # type: ignore
 
 
+class FastEmbedEmbeddings(EmbeddingModel):
+    def __init__(self, config: FastEmbedEmbeddingsConfig = FastEmbedEmbeddingsConfig()):
+        try:
+            from fastembed import TextEmbedding
+        except ImportError:
+            raise LangroidImportError("fastembed", extra="fastembed")
+
+        super().__init__()
+        self.config = config
+        self._batch_size = config.batch_size
+        self._parallel = config.parallel
+
+        self._model = TextEmbedding(
+            model_name=self.config.model_name,
+            cache_dir=self.config.cache_dir,
+            threads=self.config.threads,
+            **self.config.additional_kwargs,
+        )
+
+    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
+
+    @cached_property
+    def embedding_dims(self) -> int:
+        embed_func = self.embedding_fn()
+        return len(embed_func(["text"])[0])
+
+
 def embedding_model(embedding_fn_type: str = "openai") -> EmbeddingModel:
     """
     Args:
@@ -198,5 +249,7 @@ def embedding_model(embedding_fn_type: str = "openai") -> EmbeddingModel:
     """
     if embedding_fn_type == "openai":
         return OpenAIEmbeddings  # type: ignore
+    elif embedding_fn_type == "fastembed":
+        return FastEmbedEmbeddings  # type: ignore
     else:  # default sentence transformer
         return SentenceTransformerEmbeddings  # type: ignore

{langroid-0.2.11 → langroid-0.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "langroid"
-version = "0.2.11"
+version = "0.3.0"
 description = "Harness LLMs with Multi-Agent Programming"
 authors = ["Prasad Chalasani <pchalasani@gmail.com>"]
 readme = "README.md"
@@ -10,7 +10,7 @@ include = ["pyproject.toml"]
 
 # =============== MAIN DEPS ==============
 [tool.poetry.dependencies]
-python = "^3.10"
+python = ">=3.10,<3.13"
 
 # =========== OPTIONALS ==============================
 chromadb = {version=">=0.4.21, <=0.4.23", optional=true}
@@ -39,6 +39,7 @@ pyarrow = {version="15.0.0", optional=true}
 pdfplumber = {version="^0.10.2", optional=true}
 python-docx = {version="^1.1.0", optional=true}
 scrapy = {version="^2.11.0", optional=true}
+fastembed = {version="^0.3.1", optional=true}
 
 # ====CORE================================
 pyyaml = "^6.0.1"
@@ -111,6 +112,7 @@ all = [
     "metaphor-python", "neo4j",
     "litellm",
     "chainlit", "python-socketio",
+    "fastembed"
 ]
 # more granular groupings
 lancedb = ["lancedb", "tantivy", "pyarrow"]
@@ -135,6 +137,7 @@ mkdocs = [
 ]
 meilisearch = ["meilisearch-python-sdk"]
 momento = ["momento"]
+fastembed = ["fastembed"]
 
 
 # ================= DEV DEPS =================