INDEX

Explanations

mentions of physical piles of objects or substances

oai_token-act-pair · gpt-3.5-turbo

references to "pile" or "piles" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 obser

-0.68

 warr

-0.66

oral

-0.66

 livest

-0.65

abad

-0.63

otiation

-0.62

hips

-0.62

metic

-0.60

cffffcc

-0.58

 Pione

-0.58

POSITIVE LOGITS

 piles

1.04

 piling

1.01

 pile

1.00

bags

0.95

 piled

0.84

endium

0.77

wagon

0.76

 heap

0.76

bag

0.75

aging

0.73

Activations Density 0.017%

mentions of physical piles of objects or substances

references to "pile" or "piles" in various contexts

No Comments

No Known Activations