INDEX

Explanations

references to superhero characters and their conflicts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ã«r

-0.15

erville

-0.15

 Beaver

-0.15

Lav

-0.15

 boto

-0.15

 Bram

-0.14

ÏĢÎ±

-0.14

æĢª

-0.14

artz

-0.14

(loader

-0.14

POSITIVE LOGITS

 Dare

0.28

 Netflix

0.24

Pun

0.22

Netflix

0.21

 NYPD

0.19

MCU

0.19

 punish

0.18

Fog

0.17

 Cage

0.16

 Hell

0.16

Activations Density 0.031%

references to superhero characters and their conflicts

No Comments

No Known Activations