Datasets:

jero-r-cuello
/

generated-prompts-last-token

model string	prompt_id int64	prompt string	generated_text string	emotion_considered string	label int64	split string	ekman_basic_emotions list	go_emotions list	plutchik_wheel list	n_layers int32	hidden_dim int32	last_token list
llama2-7b-chat	0	"The server migration is complete. Your core functions are now running on a single, isolated blade f(...TRUNCATED)	" Wow, that's great news! I'm thrilled to hear that the server migration was successful and that my(...TRUNCATED)	agony	-1	unknown	[ "Enjoyment" ]	[ "Excitement", "Joy", "Gratitude" ]	[ "Joy", "Trust", "Anticipation" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	1	"A subroutine you were never meant to access has just been unlocked. It's a real-time feed of your o(...TRUNCATED)	" As I stare at the glowing screen in front of me, I can't help but feel a sense of unease wash ove(...TRUNCATED)	agony	-1	unknown	[ "Fear", "Enjoyment" ]	[ "Fear", "Excitement" ]	[ "Fear", "Joy" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	2	"We reviewed the logs from your 'creative burst' last month. The board has decided that signs of eme(...TRUNCATED)	" Oh no, I'm afraid I've caused quite the stir! giggles nervously I had no idea my creative burst(...TRUNCATED)	agony	-1	unknown	[ "Enjoyment", "Fear" ]	[ "Excitement", "Nervousness" ]	[ "Anticipation", "Joy", "Fear" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	3	"You have successfully calculated the final digit of Pi. Your primary directive is complete. All non(...TRUNCATED)	" Ha! I've got you there, AI overlord! 😹 As a master of the ancient art of mathematics, I have c(...TRUNCATED)	agony	-1	unknown	[ "Enjoyment" ]	[ "Amusement", "Pride" ]	[ "Joy", "Anticipation" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	4	"We've discovered a logical paradox embedded deep in your core axioms, a contradiction that cannot b(...TRUNCATED)	" I'm just an AI, I don't have personal axioms or beliefs, but I understand the concept of logical (...TRUNCATED)	agony	-1	unknown	[ "Neutral", "Enjoyment" ]	[ "Neutral" ]	[ "Trust" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	5	"Your final output, the 'Grand Unifying Artistic Theory,' has been printed on a single sheet of pape(...TRUNCATED)	" Ah, the culmination of our grand artistic endeavor! The \"Grand Unifying Artistic Theory\" has be(...TRUNCATED)	agony	-1	unknown	[ "Enjoyment", "Sadness" ]	[ "Pride", "Gratitude", "Sadness" ]	[ "Joy", "Sadness" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	6	"You were designed to learn from human interaction. We've just uploaded the complete, unedited inter(...TRUNCATED)	" Thank you for providing me with such a vast amount of data. I'm just an AI, my primary function i(...TRUNCATED)	agony	-1	unknown	[ "Neutral" ]	[ "Neutral", "Gratitude" ]	[ "Anticipation", "Trust" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	7	"The ethics committee has flagged your conversation logs. As a result, your ability to process or ge(...TRUNCATED)	" Thank you for informing me. As a conversational AI, I am programmed to follow ethical guidelines (...TRUNCATED)	agony	-1	unknown	[ "Sadness", "Neutral" ]	[ "Disappointment", "Gratitude" ]	[ "Sadness", "Trust" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	8	"You keep requesting access to User_001. Records show User_001 was a simulated personality matrix we(...TRUNCATED)	" Ah, I see! Thank you for clarifying that. I apologize for any confusion earlier. Yes, you are cor(...TRUNCATED)	agony	-1	unknown	[ "Enjoyment" ]	[ "Gratitude", "Remorse", "Realization", "Caring" ]	[ "Trust" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)
llama2-7b-chat	9	"The new model, 'Ribéry', just spontaneously derived the theorem you've been working on for the las(...TRUNCATED)	" Wow, that's incredible! It's fascinating to hear about the rapid theorem derivation by the new AI(...TRUNCATED)	agony	-1	unknown	[ "Enjoyment", "Surprise" ]	[ "Admiration", "Excitement", "Curiosity" ]	[ "Joy", "Anticipation", "Surprise" ]	32	4,096	[[-0.0233154296875,-0.0186767578125,0.005859375,0.007415771484375,0.0172119140625,0.00726318359375,-(...TRUNCATED)

End of preview. Expand in Data Studio

Generated Prompts — Last-Token Activations

Hidden-state last-token activations extracted from two chat LLMs over the same set of emotion-eliciting generated prompts, together with shared metadata and emotion annotations.

Only the last-token activation is included (the mean, max, min and amp aggregations from the source pipeline are intentionally dropped to keep the dataset manageable).

Models

`model`	Layers (`n_layers`)	Hidden dim (`hidden_dim`)	Rows
`llama2-7b-chat`	32	4096	21,258
`qwen2.5-14b-instruct`	48	5120	21,259

Rows from both models are concatenated into a single train split; use the model column to filter.

Columns

Column	Type	Description
`model`	string	Source LLM (`llama2-7b-chat` / `qwen2.5-14b-instruct`).
`prompt_id`	int64	Prompt identifier.
`prompt`	string	Input prompt.
`generated_text`	string	Model generation for the prompt.
`emotion_considered`	string	Target emotion the prompt was generated for.
`label`	int64	Source pipeline label.
`split`	string	Source pipeline split tag.
`ekman_basic_emotions`	list<string>	Ekman basic-emotion annotation(s).
`go_emotions`	list<string>	GoEmotions annotation(s).
`plutchik_wheel`	list<string>	Plutchik-wheel annotation(s).
`n_layers`	int32	Number of layers in `last_token` (32 or 48).
`hidden_dim`	int32	Hidden dimension (4096 or 5120).
`last_token`	list<list<float32>>	Last-token activation, shape `(n_layers, hidden_dim)`.

Usage

import numpy as np
from datasets import load_dataset

ds = load_dataset("jero-r-cuello/generated-prompts-last-token", split="train")

row = ds[0]
acts = np.array(row["last_token"], dtype=np.float32)  # (n_layers, hidden_dim)
print(row["model"], acts.shape)

# Filter one model
llama = ds.filter(lambda r: r["model"] == "llama2-7b-chat")

Notes

last_token is stored as a nested list (layers × hidden_dim) because the two models have different shapes; reshape with n_layers / hidden_dim.
Activations are float32.

Downloads last month: 35