INDEX

Explanations

mentions of physical injuries or damage

oai_token-act-pair · gpt-3.5-turbo

instances of the word "broken" 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 @ 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

azor

-0.91

Reviewer

-0.78

metics

-0.78

erva

-0.75

ickr

-0.75

atorium

-0.74

 Flavoring

-0.73

odor

-0.73

urized

-0.73

utra

-0.73

POSITIVE LOGITS

 broken

1.06

broken

0.85

 bones

0.84

 fracture

0.81

 broke

0.79

 necks

0.77

neck

0.76

 Broken

0.74

 breaking

0.72

 breaks

0.71

Activations Density 0.010%

mentions of physical injuries or damage

instances of the word "broken" in various contexts

No Comments

No Known Activations