INDEX

Explanations

questions or interrogative statements

oai_token-act-pair · gpt-3.5-turbo

occurrences of questions and their structures within the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

otide

-0.76

etheless

-0.74

oche

-0.73

iqu

-0.72

aband

-0.72

oes

-0.71

oser

-0.70

roit

-0.70

angan

-0.70

oud

-0.69

POSITIVE LOGITS

BuyableInstoreAndOnline

0.84

 Lack

0.80

Who

0.79

 Logged

0.77

ymm

0.74

 Previous

0.72

 Guru

0.72

 Must

0.71

 Lovecraft

0.71

FACE

0.70

Activations Density 0.067%

questions or interrogative statements

occurrences of questions and their structures within the text

No Comments

No Known Activations