Skip to content
The Chat Archive
Human-written demonstration (OpenAssistant volunteers)· OpenAssistant (human-written) OpenAssistant Conversations (oasst2)

I'm currently training a language model who's purpose is to figure out which emoji best…

In OpenAssistant data, “assistant” turns were written by human volunteers as demonstrations. They are not model outputs.
  1. User

    I'm currently training a language model who's purpose is to figure out which emoji best fits at the end of a sentence. Your job is to label the dataset. I wish the labels to be structured like this.

    "This day feels so good" - "😊"

    Label the below sentences:

    "I love you" "Just woke up this morning and i feel really shit" "TBH i feel like you are overreacting" "Ok" "I'm feeling a little devilish today" "OMG She's soooo cute" "I actually can't belive you fell for it"

  2. Assistant

    Sure, I can label these sentences for you. Here are my suggested emoji labels:

    "I love you" - "❤️" "Just woke up this morning and I feel really shit" - "😔" "TBH I feel like you are overreacting" - "🤔" "Ok" - "👌" "I'm feeling a little devilish today" - "😈" "OMG she's soooo cute" - "😍" "I actually can't believe you fell for it" - "🙄"

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.