-
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path.env.example
93 lines (77 loc) · 4.35 KB
/
.env.example
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
#************************************************************************************
#* References the Documentation for all the settings below here: **
#* https://github.com/jparkerweb/web-augmented-generation?tab=readme-ov-file#setup **
#************************************************************************************
######################
## General Settings ##
######################
NUM_URLS=10 # Number of URLs to fetch
SEARXNG_URL=https://searx.be/ # URL of the SearXNG server
SEARXNG_URL_EXTRA_PARAMETER="key=optional_auth_key_here&language=en" # Extra parameter for SearXNG URL
SEARXNG_FORMAT=html # Format for SearXNG results (html or json)
FETCH_TIMEOUT_MS=5000 # Timeout for fetching URLs
DISABLE_SSL_VALIDATION=true # Whether to disable SSL validation
##################
## LLM Settings ##
##################
LLM_STREAM_RESPONSE=true # Whether to stream the LLM response
#------------------
#- Ollama Local --
#------------------
LLM_BASE_URL=http://localhost:11434/v1 # Base URL for the LLM API (OpenAI format)
LLM_API_KEY=ollama!!! # API key for the LLM (use 'ollama' for Ollama)
LLM_MODEL=llama3.2:1b # Model to use with the LLM API
#----------------
#- together.ai --
#----------------
# LLM_BASE_URL=https://api.together.xyz/v1 # Base URL for the LLM API (OpenAI format)
# LLM_API_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx # API key for the LLM
# LLM_MODEL=meta-llama/Llama-3.2-3B-Instruct-Turbo # Model to use with the LLM API
#--------------
#- llama.cpp --
#--------------
# LLM_BASE_URL=http://localhost:8080/v1 # Base URL for the LLM API (OpenAI format)
# LLM_API_KEY=not-needed # API key for the LLM
# LLM_MODEL=not-needed # Model to use with the LLM API
#------------------------
#- OpenRouter Endpoint --
#------------------------
# LLM_BASE_URL=https://openrouter.ai/api/v1 # Base URL for the LLM API (OpenAI format)
# LLM_API_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx # API key for the LLM
# LLM_MODEL=google/gemini-pro-1.5-exp # Model to use with the LLM API
#------------------------------
#- Google AI Studio Endpoint --
#------------------------------
# LLM_BASE_URL=https://generativelanguage.googleapis.com/v1beta/openai/ # Base URL for the LLM API (OpenAI format)
# LLM_API_KEY=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx # API key for the LLM
# LLM_MODEL=gemini-exp-1121 # Model to use with the LLM API
####################################
## Scrapted Page Content Settings ##
####################################
#---------------------------
#- Chunk Content Settings --
#---------------------------------------------------------------------------------------------
#- These settings control how content is chunked and matched against the user's query
#- to provide better context to the LLM and reduce the number of tokens used in the LLM query.
#---------------------------------------------------------------------------------------------
CHUNK_CONTENT=true
CHUNK_CONTENT_USE_HYBRID_FALLBACK=true
## The following parameters are only used by the `chunk-match` library (if CHUNK_CONTENT is set to true)
CHUNK_CONTENT_MAX_RESULTS=10
CHUNK_CONTENT_MIN_SIMILARITY=0.375
CHUNK_CONTENT_MAX_TOKEN_SIZE=500
CHUNK_CONTENT_SIMILARITY_THRESHOLD=0.4
CHUNK_CONTENT_DYNAMIC_THRESHOLD_LOWER_BOUND=0.3
CHUNK_CONTENT_DYNAMIC_THRESHOLD_UPPER_BOUND=0.5
CHUNK_CONTENT_NUM_SIMILARITY_SENTENCES_LOOKAHEAD=3
CHUNK_CONTENT_COMBINE_CHUNKS=true
CHUNK_CONTENT_COMBINE_CHUNKS_SIMILARITY_THRESHOLD=0.5
CHUNK_CONTENT_ONNX_EMBEDDING_MODEL="Xenova/all-MiniLM-L6-v2"
CHUNK_CONTENT_DTYPE="q8"
#----------------------------------------
#- Trimmed Raw Web Page Content Length --
#----------------------------------------
# If `CHUNK_CONTENT` above is set to `false`, raw page content will be truncated
# and sent to the LLM in hopes of answering the question. The value below
# determines the maximum length of the raw page content to send to the LLM.
WEB_PAGE_CONTENT_MAX_LENGTH=1000