INDEX

Explanations

references to motorcycles

oai_token-act-pair · gpt-3.5-turbo

mentions of motorcycles

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Nanto

-0.78

ablishment

-0.77

atorial

-0.75

ochond

-0.75

tre

-0.74

iary

-0.73

 Gleaming

-0.72

onne

-0.71

bos

-0.71

bread

-0.70

POSITIVE LOGITS

 motorcycles

1.28

 motorcycle

1.23

cycles

0.98

cycle

0.98

 racing

0.87

 livest

0.85

 platoon

0.84

bike

0.84

 riding

0.81

cycl

0.80

Activations Density 0.007%

references to motorcycles

mentions of motorcycles

No Comments

No Known Activations