INDEX

Explanations

specific names of institutions and locations, potentially in a legal or official context

oai_token-act-pair · gpt-3.5-turbo

geographical locations and institutional names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 inhibitors

-0.71

oldemort

-0.68

acerb

-0.64

 haun

-0.61

krit

-0.60

ursions

-0.60

annie

-0.60

inks

-0.59

ometimes

-0.59

 chords

-0.57

POSITIVE LOGITS

 Government

0.92

 Official

0.92

 Website

0.91

 Police

0.90

 Administrative

0.90

 Historical

0.88

 Regional

0.87

 Corporation

0.84

 Logo

0.84

 Department

0.84

Activations Density 0.280%

specific names of institutions and locations, potentially in a legal or official context

geographical locations and institutional names

No Comments

No Known Activations