INDEX

Explanations

information related to the White House administration

oai_token-act-pair · gpt-3.5-turbo

references to government agencies and officials

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

ructure

-0.90

epad

-0.84

bsite

-0.82

ggle

-0.80

agame

-0.80

onymous

-0.75

ck

-0.70

ignant

-0.70

phe

-0.70

bath

-0.69

POSITIVE LOGITS

 officials

0.86

 Affairs

0.79

 staffers

0.77

 lumin

0.77

 practition

0.76

Leaks

0.75

 Arabia

0.75

 Advis

0.74

 insiders

0.74

 counterterrorism

0.74

Activations Density 0.438%

information related to the White House administration

references to government agencies and officials

No Comments

No Known Activations