INDEX

Explanations

information that has been leaked or made public without authorization

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "leaked."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

oran

-0.95

annis

-0.81

vil

-0.74

rehens

-0.74

============

-0.74

aple

-0.73

ourke

-0.72

orient

-0.72

avers

-0.70

adv

-0.68

POSITIVE LOGITS

 leaked

1.29

 leaks

1.18

 leak

1.12

 leaking

1.06

ileaks

0.96

Leaks

0.96

 confidential

0.83

 leakage

0.74

 disclosures

0.74

Activations Density 0.006%

information that has been leaked or made public without authorization

occurrences of the word "leaked."

No Comments

No Known Activations