INDEX

Explanations

the word "incidents"

oai_token-act-pair · gpt-4-turbo

keywords related to various incidents or events

oai_token-act-pair · gpt-3.5-turbo

references to incidents or occurrences, particularly in a negative or problematic context

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ramid

-0.82

zyme

-0.79

ebus

-0.78

ledge

-0.74

arton

-0.71

bern

-0.70

eday

-0.68

heit

-0.68

 bargain

-0.67

reens

-0.67

POSITIVE LOGITS

 incidents

1.02

 involving

0.97

 occurrences

0.83

 perpetrated

0.77

rences

0.77

hooting

0.76

isms

0.75

 incident

0.73

idences

0.72

iques

0.69

Activations Density 0.014%

the word "incidents"

keywords related to various incidents or events

references to incidents or occurrences, particularly in a negative or problematic context

No Comments

No Known Activations