INDEX

Explanations

captions or headings in image galleries or articles

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Caption" related to media or photographs

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

gone

-0.76

ught

-0.75

awa

-0.73

ccording

-0.69

FH

-0.69

tails

-0.66

 explan

-0.65

 toget

-0.63

bye

-0.63

Mbps

-0.63

POSITIVE LOGITS

0.87

0.83

0.81

ï

0.80

0.79

 Images

0.78

0.76

Activations Density 0.020%

captions or headings in image galleries or articles

instances of the word "Caption" related to media or photographs

No Comments

No Known Activations