INDEX

Explanations

the adjective "hard" used as an intensifier to describe difficulty

oai_token-act-pair · gpt-3.5-turbo

the phrase indicating difficulty or challenge

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

umbn

-0.71

rongh

-0.70

Fac

-0.66

igate

-0.65

ussion

-0.64

umbnail

-0.62

Kings

-0.61

roma

-0.60

tur

-0.60

bey

-0.60

POSITIVE LOGITS

 enough

0.86

 partName

0.76

oooooooo

0.72

ãĥ©

0.71

 circumst

0.70

ãĥ¼ãĥĨ

0.68

coded

0.67

 Pruitt

0.66

imaru

0.66

ãĥ¬

0.66

Activations Density 0.048%

the adjective "hard" used as an intensifier to describe difficulty

the phrase indicating difficulty or challenge

No Comments

No Known Activations