-
Notifications
You must be signed in to change notification settings - Fork 328
/
Copy pathfile9-settings.json
158 lines (158 loc) · 8.23 KB
/
file9-settings.json
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
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
{
"Logging": {
"LogLevel": {
"Default": "Warning",
// Examples: how to handle logs differently by class
// "Microsoft.KernelMemory.Handlers.TextExtractionHandler": "Information",
// "Microsoft.KernelMemory.Handlers.TextPartitioningHandler": "Information",
// "Microsoft.KernelMemory.Handlers.GenerateEmbeddingsHandler": "Information",
// "Microsoft.KernelMemory.Handlers.SaveEmbeddingsHandler": "Information",
// "Microsoft.KernelMemory.DocumentStorage.AzureBlobs": "Information",
// "Microsoft.KernelMemory.Pipeline.Queue.AzureQueues": "Information",
"Microsoft.AspNetCore": "Warning"
}
},
"KernelMemory": {
"Services": {
"AzureOpenAIEmbedding": {
// "ApiKey" or "AzureIdentity"
// AzureIdentity: use automatic Entra (AAD) authentication mechanism.
// You can test locally using the AZURE_TENANT_ID, AZURE_CLIENT_ID, AZURE_CLIENT_SECRET env vars.
"Auth": "AzureIdentity",
// Optional when Auth == AzureIdentity. Leave it null to use the default.
// in which case use this to change the client audience.
"AzureIdentityAudience": null,
"Endpoint": "https://<...>.openai.azure.com/",
"APIKey": "",
"Deployment": "",
// The max number of tokens supported by model deployed
// See https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models
"MaxTokenTotal": 8191
},
"AzureOpenAIText": {
// "ApiKey" or "AzureIdentity"
// AzureIdentity: use automatic Entra (AAD) authentication mechanism.
// You can test locally using the AZURE_TENANT_ID, AZURE_CLIENT_ID, AZURE_CLIENT_SECRET env vars.
"Auth": "AzureIdentity",
// Optional when Auth == AzureIdentity. Leave it null to use the default.
// in which case use this to change the client audience.
"AzureIdentityAudience": null,
"Endpoint": "https://<...>.openai.azure.com/",
"APIKey": "",
"Deployment": "",
// The max number of tokens supported by model deployed
// See https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models
"MaxTokenTotal": 16384,
// "ChatCompletion" or "TextCompletion"
"APIType": "ChatCompletion",
"MaxRetries": 10
},
"OpenAI": {
// Name of the model used to generate text (text completion or chat completion)
"TextModel": "gpt-4o-mini",
// The max number of tokens supported by the text model.
"TextModelMaxTokenTotal": 16384,
// What type of text generation, by default autodetect using the model name.
// Possible values: "Auto", "TextCompletion", "Chat"
"TextGenerationType": "Auto",
// Name of the model used to generate text embeddings
"EmbeddingModel": "text-embedding-ada-002",
// The max number of tokens supported by the embedding model
// See https://platform.openai.com/docs/guides/embeddings/what-are-embeddings
"EmbeddingModelMaxTokenTotal": 8191,
// OpenAI API Key
"APIKey": "",
// OpenAI Organization ID (usually empty, unless you have multiple accounts on different orgs)
"OrgId": "",
// Endpoint to use. By default the system uses 'https://api.openai.com/v1'.
// Change this to use proxies or services compatible with OpenAI HTTP protocol like LM Studio.
"Endpoint": "",
// How many times to retry in case of throttling
"MaxRetries": 10
},
"LlamaSharp": {
"TextModel": {
// path to file, e.g. "llama-2-7b-chat.Q6_K.gguf"
"ModelPath": "",
// Max number of tokens supported by the model
"MaxTokenTotal": 4096
// Optional parameters
// "GpuLayerCount": 32,
},
"EmbeddingModel": {
// path to file, e.g. "nomic-embed-text-v1.5.Q8_0.gguf"
"ModelPath": "",
// Max number of tokens supported by the model
"MaxTokenTotal": 4096
// Optional parameters
// "GpuLayerCount": 32,
}
},
"AzureAIDocIntel": {
// "APIKey" or "AzureIdentity".
// AzureIdentity: use automatic Entra (AAD) authentication mechanism.
// When the service is on sovereign clouds you can use the AZURE_AUTHORITY_HOST env var to
// set the authority host. See https://learn.microsoft.com/dotnet/api/overview/azure/identity-readme
// You can test locally using the AZURE_TENANT_ID, AZURE_CLIENT_ID, AZURE_CLIENT_SECRET env vars.
"Auth": "AzureIdentity",
// Optional when Auth == AzureIdentity. Leave it null to use the default.
// When the service is on sovereign clouds, this setting might be necessary to configure Entra auth tokens.
// See https://github.com/Azure/azure-sdk-for-net/blob/main/sdk/formrecognizer/Azure.AI.FormRecognizer/src/DocumentAnalysisAudience.cs
"AzureIdentityAudience": null,
// Required when Auth == APIKey
"APIKey": "",
"Endpoint": ""
},
"AzureAISearch": {
// "ApiKey" or "AzureIdentity". For other options see <AzureAISearchConfig>.
// AzureIdentity: use automatic Entra (AAD) authentication mechanism.
// When the service is on sovereign clouds you can use the AZURE_AUTHORITY_HOST env var to
// set the authority host. See https://learn.microsoft.com/dotnet/api/overview/azure/identity-readme
// You can test locally using the AZURE_TENANT_ID, AZURE_CLIENT_ID, AZURE_CLIENT_SECRET env vars.
"Auth": "AzureIdentity",
// Optional when Auth == AzureIdentity. Leave it null to use the default.
// When the service is on sovereign clouds, this setting might be necessary to configure Entra auth tokens.
// See https://github.com/Azure/azure-sdk-for-net/blob/main/sdk/search/Azure.Search.Documents/src/SearchAudience.cs
"AzureIdentityAudience": null,
// Required when Auth == APIKey
"APIKey": "",
"Endpoint": "https://<...>",
// Hybrid search is not enabled by default. Note that when using hybrid search
// relevance scores are different, usually lower, than when using just vector search
"UseHybridSearch": false,
// Helps improve relevance score consistency for search services with multiple replicas by
// attempting to route a given request to the same replica for that session. Use this when
// favoring consistent scoring over lower latency. Can adversely affect performance.
//
// Whether to use sticky sessions, which can help getting more consistent results.
// When using sticky sessions, a best-effort attempt will be made to target the same replica set.
// Be wary that reusing the same replica repeatedly can interfere with the load balancing of
// the requests across replicas and adversely affect the performance of the search service.
//
// See https://learn.microsoft.com/rest/api/searchservice/documents/search-post?view=rest-searchservice-2024-07-01&tabs=HTTP#request-body
"UseStickySessions": false
}
},
"Retrieval": {
"SearchClient": {
// Maximum number of tokens accepted by the LLM used to generate answers.
// The number includes the tokens used for the answer, e.g. when using
// GPT4-32k, set this number to 32768.
// If the value is not set or less than one, SearchClient will use the
// max amount of tokens supported by the model in use.
"MaxAskPromptSize": -1,
// Maximum number of relevant sources to consider when generating an answer.
// The value is also used as the max number of results returned by SearchAsync
// when passing a limit less or equal to zero.
"MaxMatchesCount": 100,
// How many tokens to reserve for the answer generated by the LLM.
// E.g. if the LLM supports max 4000 tokens, and AnswerTokens is 300, then
// the prompt sent to LLM will contain max 3700 tokens, composed by
// prompt + question + grounding information retrieved from memory.
"AnswerTokens": 300,
// Text to return when the LLM cannot produce an answer.
"EmptyAnswer": "INFO NOT FOUND"
}
}
}
}