Adjust prompt, improvements

main.py

@@ -20,19 +20,22 @@ TIMEZONE_CALGARY = pytz.timezone('America/Edmonton')
 
 bot = TelegramClient('data/bot', secrets.API_ID, secrets.API_HASH).start(bot_token=secrets.API_TOKEN)
 
-CHATGPT_TEMPLATE = '''Turn this forum post into an two paragraph instagram
-caption that tells about what a member of our makerspace has made. Add the
-hashtags: #makerspace #yyc #maker #diy #calgary and several relevant to the post
-at the end. Include a sentence explaining that this was made at Calgary
-Protospace, a makerspace that's non-profit and community ran. Only say the
-member's name once.  Use no more than 1000 characters. Write in third person.
+CHATGPT_TEMPLATE = '''
+Turn this forum post into an two paragraph summary that tells about what a member of our makerspace has made.
+Add the hashtags: #makerspace #yyc #maker #diy #calgary and several relevant to the post at the end.
+Include one sentence explaining that this was made at Calgary Protospace, a makerspace that's non-profit and volunteer run.
+Only mention the member's name once.
+Keep it short, use no more than 1000 characters.
+Write in third person.
+Use neutral sounding phrases.
 
 Title: {}
 Member: {}
 Post Body:
 ```
 {}
-```'''
+```
+'''
 
 try:
     data = json.load(open('data/data.json'))
@@ -114,8 +117,8 @@ def api_chatgpt(prompt):
 
     data = dict(
         messages=thread,
-        model='gpt-4-1106-preview',
-        temperature=0.5,
+        model='gpt-3.5-turbo',  #'gpt-4-1106-preview',
+        temperature=0.8,
         user='protogram',
         max_tokens=1000,
     )
@@ -151,7 +154,7 @@ def get_portal_name_from_discourse(username):
         logging.error('Problem with getting member name: {} - {}'.format(e.__class__.__name__, str(e)))
         return False
 
-def generate_caption(topic):
+def generate_caption(topic, prev_captions='N/A'):
     title = topic['title']
 
     username = topic['post_stream']['posts'][0]['username']
@@ -167,6 +170,7 @@ def generate_caption(topic):
     filtered_lines = [line for line in lines if 'KB' not in line and 'MB' not in line]
     body = '\n'.join(filtered_lines).replace('\n\n\n', '\n\n')
 
+    #prompt = CHATGPT_TEMPLATE.format(title, member, body, prev_captions)
     prompt = CHATGPT_TEMPLATE.format(title, member, body)
 
     logging.info('Generating caption for: {}\n{}'.format(title, body))
@@ -221,11 +225,11 @@ def find_next_valid_topic_id(topic_ids):
             if data['states'][topic_id]['status'] in ['POSTED', 'ERROR']:
                 continue
         except KeyError:
-            break
-    else:  # for loop
-        return False
+            pass
 
-    return topic_id
+        return topic_id
+
+    return False
 
 async def send_data_to_admin(state):
     try:
@@ -263,6 +267,13 @@ async def process_topics():
 
         logging.info('Next valid topic ID: {}'.format(topic_id))
 
+        try:
+            prev_states = list(data['states'].values())[-5:]
+            prev_captions = '\n\n'.join([x['caption'] for x in prev_states])
+        except:
+            prev_captions = 'N/A'
+        logging.info('Previous captions:\n{}\n\n'.format(prev_captions))
+
         if topic_id not in data['states']:
             data['states'][topic_id] = dict(
                 status='NEW',
@@ -280,7 +291,7 @@ async def process_topics():
             store_data()
             continue
 
-        caption = generate_caption(topic)
+        caption = generate_caption(topic, prev_captions)
         state['caption'] = caption
 
         result = await send_data_to_admin(state)