INDEX

Explanations

phrases related to things being straightforward or direct

oai_token-act-pair · gpt-3.5-turbo

terms related to clarity and directness

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

è¦ļéĨĴ

-0.86

 Lauder

-0.77

 livest

-0.70

mur

-0.64

ĸļ

-0.63

 perman

-0.63

mble

-0.63

chief

-0.62

oret

-0.62

 Ples

-0.62

POSITIVE LOGITS

ened

1.32

away

1.30

forward

1.26

ening

1.25

eners

1.17

edge

1.14

aways

0.94

ener

0.89

enstein

0.86

FIX

0.85

Activations Density 0.033%

phrases related to things being straightforward or direct

terms related to clarity and directness

No Comments

No Known Activations