INDEX

Explanations

numeric values

oai_token-act-pair · gpt-3.5-turbo

numerical values or identifiers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Luther

-0.77

tis

-0.76

 Flor

-0.75

 Ariel

-0.73

 Venezuel

-0.70

Îµ

-0.69

 Pengu

-0.66

ģĸ

-0.66

riel

-0.65

 Alvin

-0.64

POSITIVE LOGITS

escape

0.71

ictive

0.69

iculty

0.69

hire

0.67

acia

0.67

erent

0.66

cially

0.65

ournament

0.65

ADA

0.64

enced

0.64

Activations Density 0.000%

numeric values

numerical values or identifiers

No Comments

No Known Activations