blob: daae73a1c111ce91cd2fc120182b25bde426a83e (
plain)
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
|
[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"
|
