INDEX

Explanations

references to the year 2007

oai_token-act-pair · gpt-3.5-turbo

references to the year 2000

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

ascus

-0.77

inventoryQuantity

-0.73

ibraries

-0.71

DragonMagazine

-0.71

kell

-0.70

fired

-0.70

izont

-0.69

matically

-0.68

Kin

-0.68

oustic

-0.67

POSITIVE LOGITS

1.50

1.02

0.89

0.88

 Percent

0.88

eenth

0.86

een

0.82

ees

0.79

0.78

Activations Density 0.005%

references to the year 2007

references to the year 2000

No Comments

No Known Activations