INDEX

Explanations

specific names related to nuclear plants or incidents

oai_token-act-pair · gpt-3.5-turbo

references to specific places and individuals associated with significant events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

nesday

-0.93

aints

-0.90

abies

-0.87

ace

-0.80

cific

-0.77

mingham

-0.76

icago

-0.73

ounter

-0.72

adders

-0.72

uration

-0.71

POSITIVE LOGITS

Dai

1.08

conom

0.81

VERTISEMENT

0.80

forth

0.76

 Flavoring

0.76

ept

0.74

PUT

0.74

ãĥ´

0.74

å¥³

0.73

Mub

0.72

Activations Density 0.021%

specific names related to nuclear plants or incidents

references to specific places and individuals associated with significant events

No Comments

No Known Activations