data-prep.py

import json
import os
from collections import deque

root_dir = '~/slack_messages'

target_user = "UMQN69N8M"
res = []

for dirpath, dirnames, filenames in os.walk(root_dir):
    for file_name in filenames:
        if file_name.endswith('.json'):
            file_path = os.path.join(dirpath, file_name)
            with open(file_path, 'r') as f:
                data = json.load(f)
            buffer = deque()
            last_question = ""
            for elem in data:
                try:
                    if elem['user'] != target_user and last_question and buffer:
                        combined_message = '\n'.join(buffer)
                        res.append({"messages": [{"role": "system", "content": "You are Ross. He is a product manager at Observe inc. He responds to messages from his coworkers on slack and is helpful."}, {"role": "user", "content": last_question}, {"role": "assistant", "content": combined_message}]}
)
                        buffer.clear()
                    if elem['user'] != target_user:
                        last_question = elem['blocks'][0]['elements'][0]['elements'][0]['text']
                    elif elem['user'] == target_user:
                        buffer.append(elem['blocks'][0]['elements'][0]['elements'][0]['text'])
                # Add final conversation if exists
                    if buffer and last_question:
                        combined_message = '\n'.join(buffer)
                        res.append({"messages": [{"role": "system", "content": "You are Ross is a product manager at Observe inc. He responds to messages from his coworkers on slack and is helpful."}, {"role": "user", "content": last_question}, {"role": "assistant", "content": combined_message}]}
)
                except Exception as e:
                    continue

with open('all_messages.jsonl', 'w') as outfile:
    for entry in res:
        json.dump(entry, outfile)
        outfile.write('\n')

example-slack.json

```
[
  {
    "client_msg_id": "e59a1118-8164-4eb9-9623-e4114502ba30",
    "type": "message",
    "text": "\u003c@UMQN69N8M\u003e - How are you doing?
    "user": "U01AFPBSPCKX",
    "ts": "1644956438.392969",
    "blocks": [
      {
        "type": "rich_text",
        "block_id": "euXh",
        "elements": [
          {
            "type": "rich_text_section",
            "elements": [
              {
                "type": "user",
                "user_id": "UMQN78NMB"
              },
              {
                "type": "text",
                "text": " -  How are you doing?"
              },
              {
                "type": "link",
                "url": "https://foobar.com
              }
            ]
          }
        ]
      }
    ],
    "team": "T7B732R0S"
  },
  {
    "client_msg_id": "6590516f-ff2b-4268-aef2-f7bfadd0a250",
    "type": "message",
    "text": "I'm fine and you?",
    "user": "UMQN78NMB",
    "ts": "1644957762.921829",
    "blocks": [
      {
        "type": "rich_text",
        "block_id": "1Xntl",
        "elements": [
          {
            "type": "rich_text_section",
            "elements": [
              {
                "type": "text",
                "text": "I'm fine and you?"
              }
            ]
          }
        ]
      }
    ],
    ....
]

fine-tune.py

new file openai.FineTuningJob.create(training_file="file-jo5xMgDc8pyuP3S6IC24440T", model="gpt-3.5-turbo")

openai.FineTuningJob.create(training_file="file-jo5xMgDc8pyuP3S6IC24440T", model="gpt-3.5-turbo")