INDEX

Explanations

words related to physical objects or actions

oai_token-act-pair · gpt-3.5-turbo

word forms that relate to various suffixes and endings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

IJ

-0.62

ROM

-0.61

ologne

-0.58

DX

-0.58

QR

-0.57

 JPEG

-0.57

hers

-0.56

printf

-0.55

 Mich

-0.54

POSITIVE LOGITS

terday

0.79

hower

0.69

etheless

0.68

BuyableInstoreAndOnline

0.67

ificant

0.59

hetically

0.59

avery

0.58

heses

0.57

cond

0.57

staking

0.57

Activations Density 0.211%

words related to physical objects or actions

word forms that relate to various suffixes and endings

No Comments

No Known Activations