INDEX

Explanations

words related to restriction, censorship, or prohibition

oai_token-act-pair · gpt-3.5-turbo

terms related to blocking or censorship

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ller

-0.73

lli

-0.73

brates

-0.72

brate

-0.72

llers

-0.70

gow

-0.70

rious

-0.69

ivil

-0.66

ria

-0.66

EMBER

-0.66

POSITIVE LOGITS

 blocking

0.95

buster

0.90

busters

0.88

listed

0.80

 blockers

0.80

chains

0.78

quote

0.78

aded

0.78

lights

0.78

ades

0.77

Activations Density 0.019%

words related to restriction, censorship, or prohibition

terms related to blocking or censorship

No Comments

No Known Activations