INDEX

Explanations

conditional phrases indicating willingness or intent

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Configuration

Juliushanhanhan/llama-3-8b-it-res/blocks.25.hook_resid_post

Features

65,536

Data Type

float32

Hook Name

blocks.25.hook_resid_post

Hook Layer

Architecture

gated

Context Size

1,024

Dataset

Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 wouldn

-0.17

zim

-0.16

ricks

-0.15

iffin

-0.15

jerne

-0.15

Ascii

-0.15

ertoire

-0.15

iming

-0.14

attern

-0.14

á»¯u

-0.14

POSITIVE LOGITS

 prefer

0.25

 rather

0.25

 Rather

0.21

 Prefer

0.20

Rather

0.20

rather

0.19

prefer

0.19

nt

0.17

 prefers

0.17

 like

0.17

Activations Density 0.011%

conditional phrases indicating willingness or intent

No Known Activations