INDEX

Explanations

phrases related to legal or governmental mandates

oai_token-act-pair · gpt-3.5-turbo

references to mandates in legal or policy contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

istics

-1.05

stocks

-0.91

folk

-0.81

hop

-0.81

ienne

-0.78

icult

-0.76

cock

-0.74

hops

-0.74

stock

-0.72

acus

-0.72

POSITIVE LOGITS

 mandate

0.85

 imposed

0.78

 compel

0.76

 mandated

0.75

 creep

0.75

 requirement

0.73

 mand

0.73

 mandates

0.72

 obligation

0.72

 obedience

0.72

Activations Density 0.042%

phrases related to legal or governmental mandates

references to mandates in legal or policy contexts

No Comments

No Known Activations