INDEX

Explanations

phrases related to reasoning or explaining a conclusion

oai_token-act-pair · gpt-3.5-turbo

phrases related to accountability and consequences

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

alde

-0.63

BuyableInstoreAndOnline

-0.60

ellow

-0.56

uties

-0.56

Es

-0.56

enges

-0.56

 DevOnline

-0.56

Raw

-0.55

angler

-0.54

asse

-0.53

POSITIVE LOGITS

 thereafter

1.19

etheless

1.13

 somehow

1.12

 goddamn

1.07

 nonetheless

1.01

 fucking

1.00

 furthermore

0.98

 moreover

0.97

 thereby

0.96

 nevertheless

0.96

Activations Density 1.020%

phrases related to reasoning or explaining a conclusion

phrases related to accountability and consequences

No Comments

No Known Activations