INDEX

Explanations

instances of identity and invisibility, particularly regarding socio-cultural themes

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

uts

-0.14

arde

-0.14

acles

-0.14

 inexp

-0.13

arden

-0.13

 irresponsible

-0.13

pei

-0.13

Laz

-0.13

aeda

-0.13

iag

-0.13

POSITIVE LOGITS

 hidden

0.56

-hidden

0.50

éļĲèĹı

0.49

hidden

0.48

 hiding

0.47

 invisible

0.47

 Hidden

0.45

Hidden

0.45

 concealed

0.45

hid

0.43

Activations Density 0.424%

instances of identity and invisibility, particularly regarding socio-cultural themes

No Known Activations