-
Notifications
You must be signed in to change notification settings - Fork 39
/
Copy pathconfig.yml
65 lines (59 loc) · 2.49 KB
/
config.yml
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
# --- Base Path ---
# verbose: whether to print out module details or not
# -----------------
verbose: True
# --- Data Path ---
# data_sample: input file (.jsonl format)
# content: each file should have a text, and optionally id and label columns
# -----------------
data_sample: "data/input/sample.jsonl"
# --- Generation ---
# prompt: prompt to generate 1st-level topics
# seed: seed topics (TODO: modify to fit your conception of the data)
# output: output file containing texts and their corresponding generated topics
# topic_output: md file containing all generated topics
# -----------------
generation:
prompt: "prompt/generation_1.txt"
seed: "prompt/seed_1.md"
output: "data/output/sample/generation_1.jsonl"
topic_output: "data/output/sample/generation_1.md"
# --- Refinement ---
# refining_topics: whether to refine topics or not
# prompt: prompt to merge near duplicate topics
# output: output file, with an added refined_responses columns that replaces all removed/merged topics with the corresponding new topic
# topic_output: md file containing the list of refined topics
# mapping: mapping between originally generated topics and the topic they have been merged to
# remove: whether to remove topics with low frequency (occurring <1% of the time)
refining_topics: True
refinement:
prompt: "prompt/refinement.txt"
output: "data/output/sample/refinement.jsonl"
topic_output: "data/output/sample/refinement.md"
mapping_file: "data/output/sample/refinement_mapping.json"
remove: True
# --- Subtopic Generation ---
# generate_subtopics: whether to generate subtopics or not
# prompt: prompt to generate subtopic to each high-level topic, given some example documents that correspond to that high-level topic
# output: output file containing texts and their corresponding high-level and low-level topics
# topic_output: md file containing topic tree
# ---------------------------
generate_subtopics: True
generation_2:
prompt: "prompt/generation_2.txt"
output: "data/output/sample/generation_2.jsonl"
topic_output: "data/output/sample/generation_2.md"
# --- Assignment ---
# prompt: prompt to assign topics to documents
# output: output of topic assignment
# ------------------
assignment:
prompt: "prompt/assignment.txt"
output: "data/output/sample/assignment.jsonl"
# --- Correction ---
# prompt: prompt to correct topics
# output: output of topic correction
# ------------------
correction:
prompt: "prompt/correction.txt"
output: "data/output/sample/assignment_corrected.jsonl"