-
Notifications
You must be signed in to change notification settings - Fork 9
/
Copy pathsimpleConfig.ini
75 lines (63 loc) · 3.2 KB
/
simpleConfig.ini
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
[Default]
#This is the path to the folder where you have downloaded the repository
home_dir = /home/jovyan/work/SuperKnowa-QueryCraft/
#Unique identifier of your experiment run. The name is expected to be in following format "YOURINDETIFIER_ModelName_Date"]
exp_name = exp_codellama7b_QATesting12Mar
[DataIngestion]
#Relative path (from home_dir) of csv file to be ingested in db2 table
#CSV file for Loading
#filename = ../input/datasets/people.csv
filename = input/datasets/salary.csv
#Schema name - Database
schema_name = TheHistoryofBaseball
# Table name for CSV data
table_name= dataingestion_test_13march
[ContextRetriever]
#Relative path (from home_dir) to the folder containing sqlite database dump in .sqlite format.
#You need to provide this field only if your db_type = sqlite.
#input_database_folder= input/spider/database/
input_database_folder= input/kaggleDBQA/database/
#Relative path (from home_dir) to the csv file with columns question, query, and db_id
#input_data_file=input/datasets/spider.csv
input_data_file= input/datasets/kaggleDBQASample.csv
#input_data_file = input/datasets/DB2_Context_Input.csv
#Select either of the two source databse - sqlite or db2
db_type = db2
[Finetune]
#Data collator to be used in trainer class DataCollatorForLanguageModeling, DataCollatorForSeq2Seq or DefaultDataCollator
data_collator=DataCollatorForLanguageModeling
#Base pretrained model to finetune
model_name=codellama/CodeLlama-7b-Instruct-hf
#Relative path (from home_dir) to the text file which contains prompt tempelate
prompt_file_path=input/prompts/codellama_model.txt
#Select either of the two finetuning techniques supported - LoRA or QLoRA
finetune_type=LoRA
#Absolute path to the training dataset
#train_dataset = ${Default:home_dir}input/datasets/spiderWithContext.csv
#train_dataset = ${Default:home_dir}input/datasets/kaggleDBQAWithContext.csv
train_dataset =${Default:home_dir}input/datasets/${Default:exp_name}_contextRetriever.csv
[Inference]
## hf_batch_serial vllm_batch
inference_type=vllm_batch
model_name=codellama/CodeLlama-7b-Instruct-hf
#finetuned_model = NA
finetuned_model = ${Default:home_dir}output/model/${Default:exp_name}
#input_dataset = input/datasets/exp_codellama7b_2802_validSet.csv
input_dataset = input/datasets/${Default:exp_name}_validSet.csv
[QueryCorrection]
#Absolute path of file on which you want to get an execution accuracy score
#input_dataset = ${Default:homne_dir}output/inference/exp_codellama-13b_spider_0412.csv
input_dataset = ${Default:home_dir}output/inference/${Default:exp_name}_inference.csv
[EXEvaluator]
# provide db_type to run the code sqlite or db2. By default it is set to sqlite
db_type = db2
#Relative path (from home_dir)to the folder containing database dump in .sqlite format.
input_database_folder=${Default:home_dir}input/spider/database/
#Absolute path of file on which you want to get an execution accuracy score
#input_dataset = ${Default:home_dir}output/inference/kaggleDBQASample_inference.csv
input_dataset = ${Default:home_dir}output/inference/${Default:exp_name}_inference.csv
expected_query_column=query
generated_query_column=model_op
[QueryAnalysisDashboard]
#Absolute path of file on which you want to get an execution accuracy score
folder_name =output/evalResults/