INDEX

Explanations

references to news articles or headlines

oai_token-act-pair · gpt-3.5-turbo

instances of the word 'to' in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

gling

-0.74

phas

-0.73

cientious

-0.72

ssh

-0.69

ographically

-0.69

edIn

-0.68

Ö¼

-0.66

ibly

-0.65

mercial

-0.65

eson

-0.64

POSITIVE LOGITS

 Menu

0.87

 Contents

0.86

 Main

0.85

Top

0.85

 Gallery

0.84

TOP

0.79

 Bottom

0.78

 PAGE

0.77

 Table

0.77

 grid

0.74

Activations Density 0.042%

references to news articles or headlines

instances of the word 'to' in various contexts

No Comments

No Known Activations