INDEX

Explanations

terms related to scandals or controversies

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the word "scandal."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

guyen

-0.79

ignty

-0.75

ignt

-0.72

icrobial

-0.70

berman

-0.70

ramer

-0.70

orney

-0.69

erd

-0.68

efficients

-0.68

srf

-0.68

POSITIVE LOGITS

 scandal

1.02

 scandals

1.00

ously

0.88

 involving

0.77

icity

0.77

 revolving

0.75

 allegations

0.72

ous

0.71

ional

0.70

 Bunny

0.68

Activations Density 0.011%

terms related to scandals or controversies

repeated mentions of the word "scandal."

No Comments

No Known Activations