INDEX

Explanations

mentions of specific official titles or positions, particularly those related to law or governance

oai_token-act-pair · gpt-3.5-turbo

references to government or organizational titles and positions

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

 brace

-0.77

 track

-0.74

 flick

-0.72

 invite

-0.69

 tracks

-0.68

 drib

-0.66

 threaded

-0.64

ads

-0.63

 tracking

-0.63

 envy

-0.63

POSITIVE LOGITS

General

3.93

general

2.82

 General

2.04

 GENERAL

2.03

 general

1.55

Major

1.23

GEN

1.22

Detailed

1.21

Generic

1.20

Minor

1.19

Activations Density 0.015%

mentions of specific official titles or positions, particularly those related to law or governance

references to government or organizational titles and positions

No Comments

No Known Activations