INDEX

Explanations

references to Nazi Germany

oai_token-act-pair · gpt-3.5-turbo

references to the Nazi regime

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

trak

-1.08

tis

-1.08

tar

-0.83

ipt

-0.83

orative

-0.81

region

-0.80

-0.75

ately

-0.74

portion

-0.74

ORY

-0.74

POSITIVE LOGITS

 Nazi

0.97

 Reich

0.90

 swast

0.87

 salute

0.86

 Hitler

0.82

 Galile

0.82

 Nazis

0.79

 Adolf

0.79

 sympath

0.77

 fascist

0.77

Activations Density 0.006%

references to Nazi Germany

references to the Nazi regime

No Comments

No Known Activations