INDEX

Explanations

mentions of legal regulation or rules

oai_token-act-pair · gpt-3.5-turbo

references to regulation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 cast

-0.74

hak

-0.67

ribe

-0.66

 Pear

-0.66

atz

-0.64

nice

-0.64

 Delicious

-0.63

Za

-0.63

vik

-0.62

 Array

-0.62

POSITIVE LOGITS

 regulation

3.87

regulation

2.64

 Regulation

2.60

 regulations

2.30

 regulating

2.04

 regulate

1.96

 deregulation

1.96

 regulated

1.92

 regulatory

1.90

 regulator

1.84

Activations Density 0.009%

mentions of legal regulation or rules

references to regulation

No Comments

No Known Activations