llm-app/templates/private_rag/app.yaml at main · pathwaycom/llm-app · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
# This YAML configuration file is used to set up and configure the Private RAG template.
# It defines various components such as data sources, language models, embedders, splitters, parsers, and retrievers.
# Each section is configured to specify how the template should process and handle data for generating responses.
# You can learn more about the YAML syntax here: https://pathway.com/developers/templates/configure-yaml


# $sources defines the data sources used to read the data which will be indexed in the RAG.
# You can learn more how to configure data sources here:
# https://pathway.com/developers/templates/yaml-examples/data-sources-examples

$sources:
  # File System connector, reading data locally.
  - !pw.io.fs.read
    path: data
    format: binary
    with_metadata: true

  # Uncomment to use the SharePoint connector
  # - !pw.xpacks.connectors.sharepoint.read
  #   url: $SHAREPOINT_URL
  #   tenant: $SHAREPOINT_TENANT
  #   client_id: $SHAREPOINT_CLIENT_ID
  #   cert_path: sharepointcert.pem
  #   thumbprint: $SHAREPOINT_THUMBPRINT
  #   root_path: $SHAREPOINT_ROOT
  #   with_metadata: true
  #   refresh_interval: 30

  # Uncomment to use the Google Drive connector
  # - !pw.io.gdrive.read
  #   object_id: $DRIVE_ID
  #   service_user_credentials_file: gdrive_indexer.json
  #   file_name_pattern:
  #     - "*.pdf"
  #     - "*.pptx"
  #   object_size_limit: null
  #   with_metadata: true
  #   refresh_interval: 30


# Configures the LLM model settings for generating responses.
# The list of available Pathway LLM wrappers is available here:
# https://pathway.com/developers/api-docs/pathway-xpacks-llm/llms
# You can learn more about those in our documentation:
# https://pathway.com/developers/templates/rag-customization/llm-chats


$llm_model: "ollama/mistral"

$llm: !pw.xpacks.llm.llms.LiteLLMChat
  model: $llm_model
  retry_strategy: !pw.udfs.ExponentialBackoffRetryStrategy
    max_retries: 6
  cache_strategy: !pw.udfs.DefaultCache {}
  temperature: 0
  api_base: "http://localhost:11434"
  # api_base: "http://host.docker.internal:11434" # use this when you are running the app in the Docker on Mac or Windows
  async_mode: "fully_async"

$embedding_model: "avsolatorio/GIST-small-Embedding-v0"

# Specifies the embedder model for converting text into embeddings.
$embedder: !pw.xpacks.llm.embedders.SentenceTransformerEmbedder
  model: $embedding_model
  call_kwargs:
    show_progress_bar: False

# Sets up the splitter for chunking the documents.
$splitter: !pw.xpacks.llm.splitters.TokenCountSplitter
  max_tokens: 400

# Configures the parser for processing and extracting information from documents.
$parser: !pw.xpacks.llm.parsers.DoclingParser
  table_parsing_strategy: "llm"
  async_mode: "fully_async"
  chunk: false
  cache_strategy: !pw.udfs.DefaultCache {}

# Sets up the retriever factory for indexing and retrieving documents.
$retriever_factory: !pw.indexing.UsearchKnnFactory
  reserved_space: 1000
  embedder: $embedder
  metric: !pw.indexing.USearchMetricKind.COS

# Manages the storage and retrieval of documents for the RAG template.
$document_store: !pw.xpacks.llm.document_store.DocumentStore
  docs: $sources
  parser: $parser
  splitter: $splitter
  retriever_factory: $retriever_factory

# Configures the question-answering component using the RAG approach.
question_answerer: !pw.xpacks.llm.question_answering.AdaptiveRAGQuestionAnswerer
  llm: $llm
  indexer: $document_store
  n_starting_documents: 2
  factor: 2
  max_iterations: 4
  strict_prompt: true

# Change host and port of the webserver by uncommenting these lines
# host: "0.0.0.0"
# port: 8000

# By default, caching is enabled for UDFs with cache_strategy set.
# You can disable it by uncommenting the following line.
# persistence_mode: null
# You can also set persistence_mode to !pw.PersistenceMode.PERSISTING to enable persistence
# across restarts.
# By default, when enabled, Cache is stored in .Cache directory.
# You can customize the location by uncommenting and modifying the following lines:
# persistence_backend: !pw.persistence.Backend.filesystem
#   path: ".Cache"

# If `terminate_on_error` is true then the program will terminate whenever any error is encountered.
# Defaults to false, uncomment the following line if you want to set it to true
# terminate_on_error: true