INDEX

Explanations

phrases related to imprisonment or captives

oai_token-act-pair · gpt-3.5-turbo

references to prisoners and their conditions

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

opic

-0.79

orp

-0.73

rous

-0.70

ulously

-0.66

ript

-0.66

ories

-0.66

rious

-0.65

orie

-0.62

pered

-0.62

MN

-0.62

POSITIVE LOGITS

 prisoners

1.24

 inmates

1.07

 prisoner

1.04

 captives

1.03

 detainees

0.91

 Prison

0.86

 inmate

0.84

 hostages

0.83

 prisons

0.82

prison

0.80

Activations Density 0.012%

phrases related to imprisonment or captives

references to prisoners and their conditions

No Comments

No Known Activations