Feature/dev merge rebased (#329)

* Update local_rag.mdx

* Update llms.mdx (#322)

* update the default embd. (#310)

* Feature/add agent provider (#317)

* update pipeline (#315)

* Update CONTRIBUTING.md

* Delete CONTRIBUTOR.md

* Adding agent provider

* Feature/modify get all uniq values (#325)

* refine

* update

* format

* fix

* dev merge

.gitignore

@@ -11,7 +11,8 @@ env/
 node_modules
 
 
-**/*.sqlite
-**/*.sqlite3
+**/*.sqlite*
+**/*.sqlite3*
 
-qdrant_storage/
+qdrant_storage/
+r2r/examples/data/*

pyproject.toml

@@ -59,8 +59,8 @@ eval = ["parea-ai"]
 ionic = ["ionic-api-sdk"]
 reducto = ["boto3"]
 exa = ["exa-py"]
-sentence_transformers = ["sentence-transformers", "tokenizers"]
-local_llm = ["llama-cpp-python", "sentence-transformers", "tokenizers"]
+sentence-transformers-only = ["sentence-transformers"]
+local-llm = ["llama-cpp-python", "sentence-transformers"]
 all = ["tiktoken", "datasets", "qdrant_client", "psycopg2-binary", "sentry-sdk", "parea-ai", "boto3", "exa-py", "llama-cpp-python", "ionic-api-sdk"]
 
 [tool.poetry.group.dev.dependencies]
@@ -78,4 +78,4 @@ exclude = 'playground/.*|deprecated/.*|dump/.*|docs/source|vecs/*'
 
 [[tool.mypy.overrides]]
 module = "yaml"
-ignore_missing_imports = true
+ignore_missing_imports = true

r2r/core/__init__.py

@@ -1,5 +1,9 @@
 from .abstractions.document import DocumentPage
-from .abstractions.output import RAGPipelineOutput
+from .abstractions.output import (
+    LLMChatCompletion,
+    LLMChatCompletionChunk,
+    RAGPipelineOutput,
+)
 from .abstractions.vector import VectorEntry, VectorSearchResult
 from .agent.base import Agent
 from .pipelines.embedding import EmbeddingPipeline

r2r/core/abstractions/output.py

@@ -1,14 +1,17 @@
 from typing import Optional
 
-from openai.types.chat import ChatCompletion
+from openai.types.chat import ChatCompletion, ChatCompletionChunk
+
+LLMChatCompletion = ChatCompletion
+LLMChatCompletionChunk = ChatCompletionChunk
 
 
 class RAGPipelineOutput:
     def __init__(
         self,
         search_results: list,
         context: Optional[str] = None,
-        completion: Optional[ChatCompletion] = None,
+        completion: Optional[LLMChatCompletion] = None,
     ):
         self.search_results = search_results
         self.context = context

r2r/core/agent/base.py

@@ -4,8 +4,9 @@
 from abc import ABC, abstractmethod
 from typing import Sequence
 
+from ..abstractions.output import LLMChatCompletion
 from ..abstractions.tool import Tool
-from ..providers.llm import LLMChatCompletion, LLMProvider
+from ..providers.llm import LLMProvider
 from ..providers.prompt import PromptProvider
 
 logger = logging.getLogger(__name__)
@@ -22,6 +23,7 @@ class Agent(ABC):
     def __init__(
         self, prompt_provider: PromptProvider, llm_provider: LLMProvider
     ) -> None:
+        self.llm_provider = llm_provider
         self.prompt_provider = prompt_provider
 
         self._initialized = False

r2r/core/pipelines/rag.py

@@ -8,9 +8,8 @@
 from abc import abstractmethod
 from typing import Any, Generator, Optional, Union
 
-from openai.types.chat import ChatCompletion
 
-from ..abstractions.output import RAGPipelineOutput
+from ..abstractions.output import RAGPipelineOutput, LLMChatCompletion
 from ..providers.embedding import EmbeddingProvider
 from ..providers.llm import GenerationConfig, LLMProvider
 from ..providers.prompt import PromptProvider
@@ -120,7 +119,7 @@ def generate_completion(
         prompt: str,
         generation_config: GenerationConfig,
         conversation: list[dict] = None,
-    ) -> Union[Generator[str, None, None], ChatCompletion]:
+    ) -> Union[Generator[str, None, None], LLMChatCompletion]:
         """
         Generates a completion based on the prompt.
         """
@@ -173,7 +172,7 @@ def run(
         generation_config: Optional[GenerationConfig] = None,
         *args,
         **kwargs,
-    ) -> Union[RAGPipelineOutput, ChatCompletion]:
+    ) -> Union[RAGPipelineOutput, LLMChatCompletion]:
         """
         Runs the completion pipeline for non-streaming execution.
         """

r2r/core/providers/llm.py

@@ -4,13 +4,9 @@
 from dataclasses import dataclass, field
 from typing import List, Optional
 
-from openai.types.chat import ChatCompletion, ChatCompletionChunk
-
+from ..abstractions.output import LLMChatCompletion, LLMChatCompletionChunk
 from .base import Provider, ProviderConfig
 
-LLMChatCompletion = ChatCompletion
-LLMChatCompletionChunk = ChatCompletionChunk
-
 
 @dataclass
 class GenerationConfig(ABC):
@@ -69,7 +65,7 @@ def get_completion(
         messages: list[dict],
         generation_config: GenerationConfig,
         **kwargs,
-    ) -> ChatCompletion:
+    ) -> LLMChatCompletion:
         """Abstract method to get a chat completion from the provider."""
         pass
 
@@ -79,6 +75,6 @@ def get_completion_stream(
         messages: list[dict],
         generation_config: GenerationConfig,
         **kwargs,
-    ) -> ChatCompletionChunk:
+    ) -> LLMChatCompletionChunk:
         """Abstract method to get a completion stream from the provider."""
         pass

r2r/embeddings/setence_transformer/base.py

@@ -1,14 +1,14 @@
 import logging
 
-from r2r.core import EmbeddingProvider, VectorSearchResult
+from r2r.core import EmbeddingProvider, EmbeddingConfig, VectorSearchResult
 
 logger = logging.getLogger(__name__)
 
 
 class SentenceTransformerEmbeddingProvider(EmbeddingProvider):
     def __init__(
         self,
-        config: dict,
+        config: EmbeddingConfig,
     ):
         super().__init__(config)
         logger.info(
@@ -45,7 +45,7 @@ def __init__(
             config, EmbeddingProvider.PipelineStage.RERANK
         )
 
-    def _init_model(self, config: dict, stage: str):
+    def _init_model(self, config: EmbeddingConfig, stage: str):
         stage_name = stage.name.lower()
         model = config.get(f"{stage_name}_model", None)
         dimension = config.get(f"{stage_name}_dimension", None)

r2r/examples/clients/run_qna_client.py

@@ -9,7 +9,7 @@
 
 
 class QnAClient:
-    def __init__(self, base_url="http://localhost:8001", user_id=None):
+    def __init__(self, base_url="http://localhost:8000", user_id=None):
         self.client = R2RClient(base_url)
         if not user_id:
             self.user_id = generate_id_from_label("user_id")

r2r/examples/configs/local_llama_cpp.json

@@ -1,27 +1,27 @@
 {
   "language_model": {
-    "provider": "llama-cpp"
+    "provider": "litellm"
   },
   "vector_database": {
-    "provider": "qdrant",
+    "provider": "local",
     "collection_name": "demo_vecs"
   },
   "ingestion":{
-    "provider": "local",
-    "text_splitter": {
-      "type": "recursive_character",
-      "chunk_size": 512,
-      "chunk_overlap": 20
-    }
+    "provider": "llama-cpp"
   },
   "embedding": {
     "provider": "sentence-transformers",
     "search_model": "mixedbread-ai/mxbai-embed-large-v1",
     "search_dimension": 512,
-    "batch_size": 32
+    "batch_size": 32,
+    "text_splitter": {
+      "type": "recursive_character",
+      "chunk_size": 512,
+      "chunk_overlap": 20
+    }
   },
   "evals": {
-    "provider": "none",
+    "provider": "parea",
     "frequency": 0.0
   },
   "app": {

r2r/examples/configs/local_ollama.json

@@ -7,21 +7,21 @@
     "collection_name": "demo_vecs"
   },
   "ingestion":{
-    "provider": "local",
-    "text_splitter": {
-      "type": "recursive_character",
-      "chunk_size": 512,
-      "chunk_overlap": 20
-    }
+    "provider": "local"
   },
   "embedding": {
     "provider": "sentence-transformers",
     "search_model": "mixedbread-ai/mxbai-embed-large-v1",
     "search_dimension": 512,
-    "batch_size": 32
+    "batch_size": 32,
+    "text_splitter": {
+      "type": "recursive_character",
+      "chunk_size": 512,
+      "chunk_overlap": 20
+    }
   },
   "evals": {
-    "provider": "none",
+    "provider": "parea",
     "frequency": 0.0
   },
   "app": {

r2r/examples/configs/local_ollama_qdrant.json

@@ -7,21 +7,21 @@
     "collection_name": "demo_vecs"
   },
   "ingestion":{
-    "provider": "local",
-    "text_splitter": {
-      "type": "recursive_character",
-      "chunk_size": 512,
-      "chunk_overlap": 20
-    }
+    "provider": "local"
   },
   "embedding": {
     "provider": "sentence-transformers",
     "search_model": "mixedbread-ai/mxbai-embed-large-v1",
     "search_dimension": 512,
-    "batch_size": 32
+    "batch_size": 32,
+    "text_splitter": {
+      "type": "recursive_character",
+      "chunk_size": 512,
+      "chunk_overlap": 20
+    }
   },
   "evals": {
-    "provider": "none",
+    "provider": "parea",
     "frequency": 0.0
   },
   "app": {

r2r/examples/configs/local_ollama_with_rerank.json

@@ -1,32 +1,32 @@
 {
-  "language_model": {
-    "provider": "litellm"
-  },
-  "vector_database": {
-    "provider": "local",
-    "collection_name": "demo_vecs"
-  },
-  "ingestion":{
-    "provider": "local",
-    "text_splitter": {
-      "type": "recursive_character",
-      "chunk_size": 512,
-      "chunk_overlap": 20
-    }
-  },
   "embedding": {
     "provider": "sentence-transformers",
     "search_model": "mixedbread-ai/mxbai-embed-large-v1",
     "search_dimension": 512,
     "rerank_model": "jinaai/jina-reranker-v1-turbo-en",
     "rerank_dimension": 384,
     "rerank_transformer_type": "CrossEncoder",
-    "batch_size": 32
+    "batch_size": 32,
+    "text_splitter": {
+      "type": "recursive_character",
+      "chunk_size": 512,
+      "chunk_overlap": 20
+    }
   },
   "evals": {
     "provider": "none",
     "frequency": 0.0
   },
+  "language_model": {
+    "provider": "llama-cpp"
+  },
+  "vector_database": {
+    "provider": "qdrant",
+    "collection_name": "demo_vecs"
+  },
+  "ingestion":{
+    "provider": "local"
+  },
   "app": {
     "max_logs": 100,
     "max_file_size_in_mb": 100
@@ -36,4 +36,4 @@
     "collection_name": "demo_logs",
     "level": "INFO"
   }
-}
+}