INDEX

Explanations

in a document is locate the word "clear"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "clear" used to express clarity or obviousness 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 @ 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gins

-0.72

inqu

-0.67

izons

-0.67

rection

-0.65

inia

-0.64

rette

-0.63

inse

-0.62

illions

-0.62

umbn

-0.61

illion

-0.61

POSITIVE LOGITS

why

1.16

how

0.95

 that

0.95

why

0.92

WHY

0.87

 what

0.85

who

0.82

 enough

0.82

 whom

0.76

cut

0.76

Activations Density 0.079%

in a document is locate the word "clear"

instances of the word "clear" used to express clarity or obviousness in various contexts

No Comments

No Known Activations