INDEX

Explanations

words related to hidden information, particularly secrets

oai_token-act-pair · gpt-3.5-turbo

mentions of "secrets" and related concepts

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

allah

-0.74

olson

-0.72

à¤

-0.72

stad

-0.71

lihood

-0.70

ood

-0.70

erate

-0.68

FG

-0.67

mental

-0.66

ready

-0.65

POSITIVE LOGITS

 secrets

1.07

 Secrets

1.00

afety

0.79

 Bunny

0.78

 glean

0.76

 hidden

0.75

 Reve

0.74

cape

0.73

keeper

0.72

 unlocked

0.72

Activations Density 0.019%

words related to hidden information, particularly secrets

mentions of "secrets" and related concepts

No Comments

No Known Activations