INDEX

Explanations

the word "ultimate."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "ultimate."

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

 Cosponsors

-0.79

bread

-0.78

vere

-0.76

Ey

-0.73

 Twain

-0.72

ago

-0.71

eries

-0.71

Ey

-0.69

ep

-0.68

ald

-0.67

POSITIVE LOGITS

 ultimate

0.91

 pinnacle

0.87

 embodiment

0.87

 culmination

0.83

 guarant

0.79

 goal

0.78

 gratification

0.75

 volleyball

0.75

bledon

0.75

ierrez

0.73

Activations Density 0.004%

the word "ultimate."

instances of the word "ultimate."

No Comments

No Known Activations