INDEX

Explanations

words related to religious and ethnic groups, specifically Arabs and Jews

oai_token-act-pair · gpt-3.5-turbo

references to ethnic groups involved in conflicts, particularly Arabs and Jews

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

inventoryQuantity

-0.73

TI

-0.66

Attempts

-0.62

York

-0.61

Weapons

-0.61

Delivery

-0.60

wise

-0.60

yss

-0.58

Chain

-0.58

Hur

-0.58

POSITIVE LOGITS

aurus

1.23

paces

1.21

ervative

1.06

ervatives

1.03

hip

1.03

pace

0.97

']

0.92

hips

0.91

0.89

cale

0.88

Activations Density 0.050%

words related to religious and ethnic groups, specifically Arabs and Jews

references to ethnic groups involved in conflicts, particularly Arabs and Jews

No Comments

No Known Activations