INDEX

Explanations

words related to government agencies, particularly those involved in intelligence activities

oai_token-act-pair · gpt-3.5-turbo

mentions of intelligence committees or agencies

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

aver

-0.77

olds

-0.70

nered

-0.70

unin

-0.70

anan

-0.69

¶ħ

-0.69

interstitial

-0.69

 asphalt

-0.68

uberty

-0.68

loading

-0.68

POSITIVE LOGITS

 Agency

1.09

 Intelligence

1.05

elligence

1.00

 Surveillance

0.91

 Officer

0.84

 Investigations

0.82

 Matters

0.82

 spying

0.80

 Oversight

0.79

 Reports

0.79

Activations Density 0.012%

words related to government agencies, particularly those involved in intelligence activities

mentions of intelligence committees or agencies

No Comments

No Known Activations