INDEX

Explanations

mentions of burning or fire-related activities

oai_token-act-pair · gpt-3.5-turbo

references to the concept of burning or being burned

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

berman

-0.71

 Chomsky

-0.68

thus

-0.66

 Architects

-0.65

UFC

-0.64

atem

-0.64

plom

-0.64

Republic

-0.63

egal

-0.63

legal

-0.63

POSITIVE LOGITS

 burn

1.21

 burned

1.19

 burns

1.14

burning

1.06

 burning

1.04

 burnt

1.03

ished

0.96

 burner

0.92

burn

0.87

nesday

0.83

Activations Density 0.008%

mentions of burning or fire-related activities

references to the concept of burning or being burned

No Comments

No Known Activations