two version of R2R are here HEAD master

10 files changed, 366 insertions, 0 deletions

diff --git a/.venv/lib/python3.12/site-packages/core/configs/full.toml b/.venv/lib/python3.12/site-packages/core/configs/full.toml
new file mode 100644
index 00000000..0bf70631
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/full.toml
@@ -0,0 +1,21 @@
+[completion]
+provider = "r2r"
+concurrent_request_limit = 128
+
+[ingestion]
+provider = "unstructured_local"
+strategy = "auto"
+chunking_strategy = "by_title"
+new_after_n_chars = 2_048
+max_characters = 4_096
+combine_under_n_chars = 1_024
+overlap = 1_024
+
+    [ingestion.extra_parsers]
+    pdf = "zerox"
+
+[orchestration]
+provider = "hatchet"
+kg_creation_concurrency_limit = 32
+ingestion_concurrency_limit = 16
+kg_concurrency_limit = 8
diff --git a/.venv/lib/python3.12/site-packages/core/configs/full_azure.toml b/.venv/lib/python3.12/site-packages/core/configs/full_azure.toml
new file mode 100644
index 00000000..c6ebb199
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/full_azure.toml
@@ -0,0 +1,46 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "azure/gpt-4o-mini"
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "azure/gpt-4o"
+
+# LLM used for ingesting visual inputs
+vlm = "azure/gpt-4o"
+
+# LLM used for transcription
+audio_lm = "azure/whisper-1"
+
+# Reasoning model, used for `research` agent
+reasoning_llm = "azure/o3-mini"
+# Planning model, used for `research` agent
+planning_llm = "azure/o3-mini"
+
+[embedding]
+base_model = "azure/text-embedding-3-small"
+
+[completion_embedding]
+base_model = "azure/text-embedding-3-small"
+
+[ingestion]
+provider = "unstructured_local"
+strategy = "auto"
+chunking_strategy = "by_title"
+new_after_n_chars = 2_048
+max_characters = 4_096
+combine_under_n_chars = 1_024
+overlap = 1_024
+document_summary_model = "azure/gpt-4o-mini"
+automatic_extraction = true # enable automatic extraction of entities and relations
+
+  [ingestion.extra_parsers]
+    pdf = "zerox"
+
+  [ingestion.chunk_enrichment_settings]
+    generation_config = { model = "azure/gpt-4o-mini" }
+
+[orchestration]
+provider = "hatchet"
+kg_creation_concurrency_limit = 32
+ingestion_concurrency_limit = 4
+kg_concurrency_limit = 8
diff --git a/.venv/lib/python3.12/site-packages/core/configs/full_lm_studio.toml b/.venv/lib/python3.12/site-packages/core/configs/full_lm_studio.toml
new file mode 100644
index 00000000..daae73a1
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/full_lm_studio.toml
@@ -0,0 +1,57 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "lm_studio/llama-3.2-3b-instruct"
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "lm_studio/llama-3.2-3b-instruct"
+
+# LLM used for ingesting visual inputs
+vlm = "lm_studio/llama3.2-vision" # TODO - Replace with viable candidate
+
+# LLM used for transcription
+audio_lm = "lm_studio/llama-3.2-3b-instruct" # TODO - Replace with viable candidate
+
+[embedding]
+provider = "litellm"
+base_model = "lm_studio/text-embedding-nomic-embed-text-v1.5"
+base_dimension = nan
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[completion_embedding]
+# Generally this should be the same as the embedding config, but advanced users may want to run with a different provider to reduce latency
+provider = "litellm"
+base_model = "lm_studio/text-embedding-nomic-embed-text-v1.5"
+base_dimension = nan
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[agent]
+tools = ["search_file_knowledge"]
+
+[completion]
+provider = "litellm"
+concurrent_request_limit = 1
+
+  [completion.generation_config]
+  temperature = 0.1
+  top_p = 1
+  max_tokens_to_sample = 1_024
+  stream = false
+
+[ingestion]
+provider = "unstructured_local"
+strategy = "auto"
+chunking_strategy = "by_title"
+new_after_n_chars = 512
+max_characters = 1_024
+combine_under_n_chars = 128
+overlap = 20
+chunks_for_document_summary = 16
+document_summary_model = "lm_studio/llama-3.2-3b-instruct"
+automatic_extraction = false
+
+[orchestration]
+provider = "hatchet"
diff --git a/.venv/lib/python3.12/site-packages/core/configs/full_ollama.toml b/.venv/lib/python3.12/site-packages/core/configs/full_ollama.toml
new file mode 100644
index 00000000..8ec2fc77
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/full_ollama.toml
@@ -0,0 +1,63 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "openai/llama3.1"
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "openai/llama3.1"
+
+# LLM used for ingesting visual inputs
+vlm = "openai/llama3.1" # TODO - Replace with viable candidate
+
+# LLM used for transcription
+audio_lm = "openai/llama3.1" # TODO - Replace with viable candidate
+
+
+# Reasoning model, used for `research` agent
+reasoning_llm = "openai/llama3.1"
+# Planning model, used for `research` agent
+planning_llm = "openai/llama3.1"
+
+[embedding]
+provider = "ollama"
+base_model = "mxbai-embed-large"
+base_dimension = 1_024
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[completion_embedding]
+provider = "ollama"
+base_model = "mxbai-embed-large"
+base_dimension = 1_024
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[agent]
+tools = ["search_file_knowledge"]
+
+[completion]
+provider = "litellm"
+concurrent_request_limit = 1
+
+  [completion.generation_config]
+  temperature = 0.1
+  top_p = 1
+  max_tokens_to_sample = 1_024
+  stream = false
+  api_base = "http://localhost:11434/v1"
+
+[ingestion]
+provider = "unstructured_local"
+strategy = "auto"
+chunking_strategy = "by_title"
+new_after_n_chars = 512
+max_characters = 1_024
+combine_under_n_chars = 128
+overlap = 20
+chunks_for_document_summary = 16
+document_summary_model = "ollama/llama3.1"
+automatic_extraction = false
+
+[orchestration]
+provider = "hatchet"
diff --git a/.venv/lib/python3.12/site-packages/core/configs/gemini.toml b/.venv/lib/python3.12/site-packages/core/configs/gemini.toml
new file mode 100644
index 00000000..50739a6c
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/gemini.toml
@@ -0,0 +1,21 @@
+[app]
+fast_llm = "gemini/gemini-2.0-flash-lite"
+quality_llm = "gemini/gemini-2.0-flash"
+vlm = "gemini/gemini-2.0-flash"
+audio_lm = "gemini/gemini-2.0-flash-lite"
+
+[embedding]
+provider = "litellm"
+base_model = "gemini/text-embedding-004"
+base_dimension = nan
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[completion_embedding]
+provider = "litellm"
+base_model = "gemini/text-embedding-004"
+base_dimension = nan
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
diff --git a/.venv/lib/python3.12/site-packages/core/configs/lm_studio.toml b/.venv/lib/python3.12/site-packages/core/configs/lm_studio.toml
new file mode 100644
index 00000000..1b8acb8f
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/lm_studio.toml
@@ -0,0 +1,42 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "lm_studio/llama-3.2-3b-instruct"
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "lm_studio/llama-3.2-3b-instruct"
+
+# LLM used for ingesting visual inputs
+vlm = "lm_studio/llama3.2-vision" # TODO - Replace with viable candidate
+
+# LLM used for transcription
+audio_lm = "lm_studio/llama-3.2-3b-instruct" # TODO - Replace with viable candidate
+
+[embedding]
+provider = "litellm"
+base_model = "lm_studio/text-embedding-nomic-embed-text-v1.5"
+base_dimension = nan
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[completion_embedding]
+# Generally this should be the same as the embedding config, but advanced users may want to run with a different provider to reduce latency
+provider = "litellm"
+base_model = "lm_studio/text-embedding-nomic-embed-text-v1.5"
+base_dimension = nan
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[agent]
+tools = ["search_file_knowledge"]
+
+[completion]
+provider = "litellm"
+concurrent_request_limit = 1
+
+  [completion.generation_config]
+  temperature = 0.1
+  top_p = 1
+  max_tokens_to_sample = 1_024
+  stream = false
diff --git a/.venv/lib/python3.12/site-packages/core/configs/ollama.toml b/.venv/lib/python3.12/site-packages/core/configs/ollama.toml
new file mode 100644
index 00000000..5226eebf
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/ollama.toml
@@ -0,0 +1,48 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "openai/llama3.1" ### NOTE - RECOMMENDED TO USE `openai` with `api_base = "http://localhost:11434/v1"` for best results, otherwise `ollama` with `litellm` is acceptable
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "openai/llama3.1"
+
+# LLM used for ingesting visual inputs
+vlm = "openai/llama3.1" # TODO - Replace with viable candidate
+
+# LLM used for transcription
+audio_lm = "openai/llama3.1" # TODO - Replace with viable candidate
+
+
+# Reasoning model, used for `research` agent
+reasoning_llm = "openai/llama3.1"
+# Planning model, used for `research` agent
+planning_llm = "openai/llama3.1"
+
+[embedding]
+provider = "ollama"
+base_model = "mxbai-embed-large"
+base_dimension = 1_024
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[completion_embedding]
+provider = "ollama"
+base_model = "mxbai-embed-large"
+base_dimension = 1_024
+batch_size = 128
+add_title_as_prefix = true
+concurrent_request_limit = 2
+
+[agent]
+tools = ["search_file_knowledge"]
+
+[completion]
+provider = "litellm"
+concurrent_request_limit = 1
+
+  [completion.generation_config]
+  temperature = 0.1
+  top_p = 1
+  max_tokens_to_sample = 1_024
+  stream = false
+  api_base = "http://localhost:11434/v1"
diff --git a/.venv/lib/python3.12/site-packages/core/configs/r2r_azure.toml b/.venv/lib/python3.12/site-packages/core/configs/r2r_azure.toml
new file mode 100644
index 00000000..fec2b026
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/r2r_azure.toml
@@ -0,0 +1,23 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "azure/gpt-4o-mini"
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "azure/gpt-4o"
+
+# LLM used for ingesting visual inputs
+vlm = "azure/gpt-4o"
+
+# LLM used for transcription
+audio_lm = "azure/whisper-1"
+
+# Reasoning model, used for `research` agent
+reasoning_llm = "azure/o3-mini"
+# Planning model, used for `research` agent
+planning_llm = "azure/o3-mini"
+
+[embedding]
+base_model = "azure/text-embedding-3-small"
+
+[completion_embedding]
+base_model = "azure/text-embedding-3-small"
diff --git a/.venv/lib/python3.12/site-packages/core/configs/r2r_azure_with_test_limits.toml b/.venv/lib/python3.12/site-packages/core/configs/r2r_azure_with_test_limits.toml
new file mode 100644
index 00000000..d26e7683
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/r2r_azure_with_test_limits.toml
@@ -0,0 +1,37 @@
+[app]
+# LLM used for internal operations, like deriving conversation names
+fast_llm = "azure/gpt-4o-mini"
+
+# LLM used for user-facing output, like RAG replies
+quality_llm = "azure/gpt-4o"
+
+# LLM used for ingesting visual inputs
+vlm = "azure/gpt-4o"
+
+# LLM used for transcription
+audio_lm = "azure/whisper-1"
+
+
+# Reasoning model, used for `research` agent
+reasoning_llm = "azure/o3-mini"
+# Planning model, used for `research` agent
+planning_llm = "azure/o3-mini"
+
+[embedding]
+base_model = "azure/text-embedding-3-small"
+base_dimension = 512
+
+[completion_embedding]
+base_model = "azure/text-embedding-3-small"
+
+[database]
+  [database.limits]
+  global_per_min = 10  # Small enough to test quickly
+  monthly_limit = 20  # Small enough to test in one run
+
+  [database.route_limits]
+  "/v3/retrieval/search" = { route_per_min = 5, monthly_limit = 10 }
+
+  [database.user_limits."47e53676-b478-5b3f-a409-234ca2164de5"]
+  global_per_min = 2
+  route_per_min = 1
diff --git a/.venv/lib/python3.12/site-packages/core/configs/r2r_with_auth.toml b/.venv/lib/python3.12/site-packages/core/configs/r2r_with_auth.toml
new file mode 100644
index 00000000..f36e8bb3
--- /dev/null
+++ b/.venv/lib/python3.12/site-packages/core/configs/r2r_with_auth.toml
@@ -0,0 +1,8 @@
+[auth]
+provider = "r2r"
+access_token_lifetime_in_minutes = 60
+refresh_token_lifetime_in_days = 7
+require_authentication = true
+require_email_verification = false
+default_admin_email = "admin@example.com"
+default_admin_password = "change_me_immediately"