NC Bench
Overview
About
Tests
Models
Model
Hermes 2 Theta 8B
Test
Dialogue tags
Scenario
Write 500 words with 30% dialogue
Run ID
2024-07-19-01J35SPGKTAHR03PVF4NPJWXDX
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% words of dialogue between two thieves named Molly and Bill who are planning to break into a children's party to steal all the presents. Only include the prose of the scene without any markdown or commentary.
Model response
Messages
Meta
AI
The sun was setting over the suburban neighbourhood, casting a warm orange glow over the scene before them. Molly and Bill stood side-by-side, dressed in their finest civilian disguises, surveying the target from across the street. A garden of colour and laughter, balloons bobbing in the breeze, excited squeals of children echoed through the evening air. "Our objective," Bill whispered, "is to break into said party-house, acquire all gifts, and make a swift, clean exit without alerting the security guards in the form of a seven-year-old in a blue cape, who's currently bouncing around like a hyperactive human pinball." Molly nodded, her eyes twinkling with mischief. "The hundo-pardo. I've got a plan." Bill's eyes rolled skyward. "Oh boy, do tell." Molly explained their strategy, which involved using a hose to slip under the back fence, disguising themselves as entertainers, and exploiting the chaos of a piñata-breaking session to swap out presents with identical-looking dog toys. Bill listened, expression neutral, until she reached the 'identical-looking dog toys' part. "You've thought this through, haven't you, Molly?" he asked, a hint of disbelief creeping into his voice. "Just consider the odds," she replied mysteriously. "Even if the Little Timmy's or whoever recognizes one of our more, ah, 'presents', the chaos will be so immense that by the time someone realizes something's amiss, we'll be long gone with a bagful of hot toys to sell." Bill sighed. "Alright, alright. Plan C, let's do it. But, uh, you owe me for this." As dusk fell, they crept into the garden, Molly pulling the hose while Bill used it to slip under the fence. Once inside, they exchanged their suits for matching superhero outfits and red wigs – stolen from a previous heist, Bill reminded her. They entered the party, Molly clanging two cymbals together to announce their arrival, while Bill held back, disguised as a balloon-animal twister. The party-goers cheered as Molly commenced her routine, juggling beach balls and fire-engine red candy apples with incredible skill. A seven-year-old with a mohawk ran up, interested in a demonstration of her fire-eating skills. Molly winked at Bill, Ingrid Bergman-style, and produced a small canister that emitted a lifelike fire-breath simulation. The boy's eyes went wide, and he passed along the word. Security guard number one, the aforementioned blue-caped vigilante, swooped in, demanding an autograph for his cousin. Bill handed over a piece of paper and continued inflating balloons while guiding the boy's attention to the piñata. As Molly performed her final trick – a spectacular juggling act that fell prey to gravity – Bill discreetly swapped out the gifts. It was almost elegant. Simple, but elegant. They slipped away, their bags bulging with loot as the chaos of the piñata-breaking session erupted. The now-empty boxes replaced with identical dog toys that the kids barely blinked at. And then they were gone, the shadows swallowing them as they vanished into the night, their laughter echoing behind them like a promise of another successful heist.
id
"gen-dWBf68BHkEd4BHc7XHWDioJvH1Nc"
total_cost
0.00077
upstream_id
"5f8e9b281SPbZZo6"
created_at
"2024-07-19T15:28:09.07441+00:00"
Friday, July 19th, 2024 at 3:28:09 PM GMT+00:00
model
"nousresearch/hermes-2-theta-llama-3-8b"
app_id
182717
streamed
false
cancelled
false
provider_name
"Mancer"
latency
20442
moderation_latency
(null)
generation_time
0
finish_reason
"stop"
tokens_prompt
67
tokens_completion
674
native_tokens_prompt
70
native_tokens_completion
674
num_media_prompt
(null)
num_media_completion
(null)
origin
"https://ncbench.com/"
usage
0.00077
Evaluation details
Result
Evaluator
Details
Meta Data
75.1556%
Matches word count
n/a
neededClean
false
words
513
20.4573%
Dialogue to Total Word Ratio
Ratio: 24.37%, Deviation: 5.63%
neededClean
false
wordsTotal
513
wordsDialogue
125
47.8064%