INDEX

Explanations

words related to forceful actions or transformations

oai_token-act-pair · gpt-3.5-turbo

words that end in the suffix 'ate' or variations of it

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

bang

-0.72

REL

-0.72

amar

-0.70

owitz

-0.70

azaki

-0.69

follow

-0.66

swick

-0.66

drawn

-0.65

backs

-0.65

é¾įå

-0.63

POSITIVE LOGITS

urally

0.74

 libel

0.73

ulate

0.72

mble

0.72

iate

0.72

ATING

0.72

ATED

0.69

ly

0.68

uate

0.65

aneously

0.65

Activations Density 0.229%

words related to forceful actions or transformations

words that end in the suffix 'ate' or variations of it

No Comments

No Known Activations