INDEX

Explanations

words related to specific individuals and institutions, particularly those named Andrews

oai_token-act-pair · gpt-3.5-turbo

names and references to specific people and places

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

pher

-0.88

kered

-0.80

osph

-0.80

orescence

-0.78

uana

-0.76

aptic

-0.74

icides

-0.73

©¶æ

-0.72

ensing

-0.72

orescent

-0.71

POSITIVE LOGITS

shire

0.85

 Annex

0.83

 Swim

0.79

mire

0.74

 Dull

0.74

 contractor

0.72

gallery

0.72

landish

0.71

DragonMagazine

0.66

 Takeru

0.66

Activations Density 0.041%

words related to specific individuals and institutions, particularly those named Andrews

names and references to specific people and places

No Comments

No Known Activations