INDEX

Explanations

keywords related to leaked or classified information, potentially from large scale information releases such as WikiLeaks

oai_token-act-pair · gpt-3.5-turbo

terms related to leaking and exposing information or documents

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

IFrame

Link

Negative Logits

stad

-0.73

 neigh

-0.70

nir

-0.70

Gau

-0.69

inis

-0.69

allic

-0.68

vic

-0.68

 XIII

-0.67

inence

-0.67

ural

-0.64

POSITIVE LOGITS

 trove

1.31

 WikiLeaks

1.22

 whistlebl

1.18

 Snowden

1.11

 revelations

1.10

 whistleblower

1.09

 Wikileaks

1.08

 disclosures

1.07

ileaks

1.07

Leaks

1.06

Activations Density 0.138%

keywords related to leaked or classified information, potentially from large scale information releases such as WikiLeaks

terms related to leaking and exposing information or documents

No Known Activations