INDEX

Explanations

rankings or positions indicated by the phrase "No." followed by a number

oai_token-act-pair · gpt-3.5-turbo

rankings or positions associated with numerical values in a competitive context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

theless

-0.75

 totality

-0.65

 disg

-0.64

redits

-0.63

staking

-0.61

endish

-0.60

 muted

-0.59

tones

-0.59

 staging

-0.56

liament

-0.56

POSITIVE LOGITS

xious

1.02

onday

0.78

osphere

0.76

obs

0.76

isy

0.75

Va

0.74

zzle

0.74

DIV

0.72

pper

0.70

Ã«

0.70

Activations Density 0.019%

rankings or positions indicated by the phrase "No." followed by a number

rankings or positions associated with numerical values in a competitive context

No Comments

No Known Activations