INDEX

Explanations

words related to complex and intricate structures, such as labyrinths

oai_token-act-pair · gpt-3.5-turbo

thematic elements related to complex narratives and relationships

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ĵĺ

-0.75

cki

-0.69

velength

-0.68

llular

-0.65

avan

-0.65

rounder

-0.64

ploy

-0.64

 retirees

-0.64

agh

-0.63

VO

-0.62

POSITIVE LOGITS

yrinth

0.99

inates

0.93

inators

0.85

ination

0.84

ics

0.83

inatory

0.83

 Artemis

0.83

inations

0.81

 Cinderella

0.81

inator

0.80

Activations Density 0.031%

words related to complex and intricate structures, such as labyrinths

thematic elements related to complex narratives and relationships

No Comments

No Known Activations