INDEX

Explanations

phrases indicating comparison or contrast

oai_token-act-pair · gpt-3.5-turbo

conditional phrases and implications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ouble

-0.68

OGR

-0.67

voc

-0.67

ãĤ¨ãĥ«

-0.66

ThumbnailImage

-0.65

Detailed

-0.65

tained

-0.65

VERTISEMENT

-0.64

pec

-0.64

alysed

-0.64

POSITIVE LOGITS

then

1.18

 then

1.10

 surely

1.01

 chances

0.93

 THEN

0.93

why

0.91

it

0.78

 congratulations

0.78

 please

0.75

 maybe

0.72

Activations Density 0.228%

phrases indicating comparison or contrast

conditional phrases and implications

No Comments

No Known Activations