RAG-training/config.yml at main · SeaseLtd/RAG-training · 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
# Solr url where the data is indexed
solr_url: http://localhost:8984/solr/rag_index

# Server port where the langchain server runs
server_port: 8000

# If you set  allenai/scibert_scivocab_uncased it is going to download it at runtime.
# If you have already downloaded the model, you can set here the local path
pretrained_model_path: allenai/scibert_scivocab_uncased

# Min size (number of characters) of a chunk to be indexed
# All the chunks with fewer characters will be discarded
min_chunk_size: 500

# RAG endpoint
# All the fields are mandatory
endpoint:
  -
    # OpenAI model name
    model_name: "gpt-3.5-turbo-1106"
    # The name of the endpoint. The server will be run at the address http://URL:<server_port>/<endpoint_name>/playground
    endpoint_name: "gpt_3_5_rag"
    # Number of rows to retrieve
    rows: 10