INDEX

Explanations

names of people, particularly "Goldberg."

oai_token-act-pair · gpt-3.5-turbo

mentions of specific names and political correctness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hur

-0.73

ampa

-0.65

ewater

-0.63

emer

-0.62

Ey

-0.60

TIT

-0.60

arb

-0.60

 Mare

-0.60

mir

-0.59

hd

-0.59

POSITIVE LOGITS

 Goldberg

3.02

 correctness

2.09

 Gupta

1.18

 Jinping

1.16

sq

0.97

aque

0.91

instein

0.88

 Enix

0.87

 suprem

0.82

 Ellison

0.81

Activations Density 0.028%

names of people, particularly "Goldberg."

mentions of specific names and political correctness

No Comments

No Known Activations