initial commit

This commit is contained in:
wea_ondara
2024-04-17 18:58:50 +02:00
commit 41d5b4f591
22 changed files with 1611 additions and 0 deletions

View File

@@ -0,0 +1,27 @@
import json
import os
this_dir = os.path.dirname(os.path.abspath(__file__))
def mkdir(path):
if not os.path.isdir(path):
os.mkdir(path)
mkdir(this_dir + '/../../data')
mkdir(this_dir + '/../../data/helpsteer')
for filename in ['train.jsonl', 'validation.jsonl']:
with open(this_dir + '/' + filename, 'r') as f:
lines = f.readlines()
role_dict = {'prompt': 'user', 'response': 'assistant'}
lines = [json.loads(line) for line in lines]
conversations = [{'messages': [{'role': 'user', 'content': line['prompt']},
{'role': 'assistant', 'content': line['response']}]} for line in lines]
print(conversations[0])
with open(this_dir + '/../../data/helpsteer/helpsteer_' + filename[0:-6] + '_all.jsonl', 'w') as f:
f.writelines([json.dumps(conv) + '\n' for conv in conversations])

View File

@@ -0,0 +1,42 @@
import json
import os
# parsing OA data files with oasst_data helpers
from oasst_data import read_message_trees, ExportMessageNode
messages: list[ExportMessageNode] = []
this_dir = os.path.dirname(os.path.abspath(__file__))
input_file_path = this_dir + '/2023-11-05_oasst2_all.trees.jsonl.gz'
role_dict = {'prompter': 'user', 'assistant': 'assistant'}
conversations = []
def visit(node: ExportMessageNode, parents: [ExportMessageNode]):
new_parents = parents + [node]
if not node.replies: # end of conversation
conversations.append({'messages': [{'role': role_dict[p.role], 'content': p.text} for p in new_parents]})
else:
for reply in node.replies:
visit(reply, new_parents)
for tree in read_message_trees(input_file_path):
if tree.prompt.lang not in ['en']: # filtering by language tag (optional)
continue
visit(tree.prompt, [])
print(conversations[0])
def mkdir(path):
if not os.path.isdir(path):
os.mkdir(path)
mkdir(this_dir + '/../../data')
mkdir(this_dir + '/../../data/oasst')
with open(this_dir + '/../../data/oasst/oasst_all.jsonl', 'w') as f:
f.writelines([json.dumps(conv) + '\n' for conv in conversations])