Initial Query

For each of the following ten movies give at most 5 emoji that would best describe the movie. Give your answer as a JSON array. So If I asked for 

1
you might might answer 
1
2
{"Finding Nemo": ["🐠", "🐟", "🐡", "🐬", "🐳"],
"Toy Story": ["🚀", "⚔️,", "🤖", "👽", "🌌"]}
. Each emoji must be a single utf8 character. ABSOLUTELY NO ZERO WIDTH JOINING. (So, max(len(emoji) for movie in result.values() for emoji in movie) == 1) Now give me answers for these movies: 
1
Give ONLY a JSON output. Nothing else.

LLM Generation

Query

For each of the following ten movies give at most 5 emoji that would best describe the movie. Give your answer as a JSON array. So If I asked for 

1
you might might answer 
1
2
{"Finding Nemo": ["🐠", "🐟", "🐡", "🐬", "🐳"],
"Toy Story": ["🚀", "⚔️,", "🤖", "👽", "🌌"]}
. Each emoji must be a single utf8 character. ABSOLUTELY NO ZERO WIDTH JOINING. (So, max(len(emoji) for movie in result.values() for emoji in movie) == 1) Now give me answers for these movies: 
1
Give ONLY a JSON output. Nothing else.

Output

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
{
  "The Lion King": ["🦁", "👑", "🌍", "🎶", "☀️"],
  "The Nightmare Before Christmas": ["💀", "🎄", "🎃", "👻", "🚪"],
  "The Godfather": ["🤵", "🔫", "💰", "🍊", "🇮🇹"],
  "The Matrix": ["💻", "💊", "🕶️", "☎️", "🥄"],
  "Casablanca": ["✈️", "🎹", "🎩", "💔", "😢"],
  "Raiders of the Lost Ark": ["🤠", "🐍", "💎", "🏺", "💰"],
  "V for Vendetta": ["🎭", "🔪", "💥", "🌹", "🏛️"],
  "The Princess Bride": ["⚔️", "👑", "☠️", "💖", "📖"],
  "Back to the Future": ["🚗", "⏰", "⚡", "🎸", "🔥"],
  "Dune": ["🏜️", "🐛", "💧", "👁️", "✨"]
}

Extract Json

I extracted the following JSON from that output: 

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
{
  "The Lion King": ["🦁", "👑", "🌍", "🎶", "☀️"],
  "The Nightmare Before Christmas": ["💀", "🎄", "🎃", "👻", "🚪"],
  "The Godfather": ["🤵", "🔫", "💰", "🍊", "🇮🇹"],
  "The Matrix": ["💻", "💊", "🕶️", "☎️", "🥄"],
  "Casablanca": ["✈️", "🎹", "🎩", "💔", "😢"],
  "Raiders of the Lost Ark": ["🤠", "🐍", "💎", "🏺", "💰"],
  "V for Vendetta": ["🎭", "🔪", "💥", "🌹", "🏛️"],
  "The Princess Bride": ["⚔️", "👑", "☠️", "💖", "📖"],
  "Back to the Future": ["🚗", "⏰", "⚡", "🎸", "🔥"],
  "Dune": ["🏜️", "🐛", "💧", "👁️", "✨"]
}

PyFunc

Resulting in output: [['🦁', '👑', '🌍', '🎶', '☀️'], ['💀', '🎄', '🎃', '👻', '🚪'], ['🤵', '🔫', '💰', '🍊', '🇮🇹'], ['💻', '💊', '🕶️', '☎️', '🥄'], ['✈️', '🎹', '🎩', '💔', '😢'], ['🤠', '🐍', '💎', '🏺', '💰'], ['🎭', '🔪', '💥', '🌹', '🏛️'], ['⚔️', '👑', '☠️', '💖', '📖'], ['🚗', '⏰', '⚡', '🎸', '🔥'], ['🏜️', '🐛', '💧', '👁️', '✨']]

LLM Generation

Query

For each of the following ten movies described by 5 emoji, give the movie title that best matches. Give your answer as a JSON list. So If I asked for 

1
You might answer 
1
["Finding Nemo", "Toy Story"]]
. Now give me answers for these movies: 
1
2
3
4
5
6
7
8
9
['💀', '🎄', '🎃', '👻', '🚪'],
['🤵', '🔫', '💰', '🍊', '🇮🇹'],
['💻', '💊', '🕶️', '☎️', '🥄'],
['✈️', '🎹', '🎩', '💔', '😢'],
['🤠', '🐍', '💎', '🏺', '💰'],
['🎭', '🔪', '💥', '🌹', '🏛️'],
['⚔️', '👑', '☠️', '💖', '📖'],
['🚗', '⏰', '⚡', '🎸', '🔥'],
['🏜️', '🐛', '💧', '👁️', '✨']]
What are the names of the movie titles?

Output

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
[
  "The Lion King",
  "The Nightmare Before Christmas",
  "The Godfather",
  "The Matrix",
  "Casablanca",
  "Raiders of the Lost Ark",
  "V for Vendetta",
  "The Princess Bride",
  "Back to the Future",
  "Dune"
]

Extract Json

I extracted the following JSON from that output: 

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
[
  "The Lion King",
  "The Nightmare Before Christmas",
  "The Godfather",
  "The Matrix",
  "Casablanca",
  "Raiders of the Lost Ark",
  "V for Vendetta",
  "The Princess Bride",
  "Back to the Future",
  "Dune"
]

PyFunc

OK Resulting in output: True