INDEX

Explanations

words related to specific entities or topics, such as "Flood" or "Tongue-in-cheek faux ads"

oai_token-act-pair · gpt-3.5-turbo

proper nouns and brand names

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

-0.52

nesota

-0.52

orney

-0.48

Downloadha

-0.46

emale

-0.46

 beginning

-0.45

},{"

-0.45

],"

-0.44

SPONSORED

-0.44

latest

-0.44

POSITIVE LOGITS

hetically

0.58

quartered

0.53

cknowled

0.50

ogether

0.50

Wiki

0.49

Wire

0.49

ishly

0.49

Gen

0.48

ually

0.48

itting

0.48

Activations Density 0.707%

words related to specific entities or topics, such as "Flood" or "Tongue-in-cheek faux ads"

proper nouns and brand names

No Comments

No Known Activations