INDEX

Explanations

phrases describing attempts to control or restrain something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "suppress" and related forms

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

psc

-0.77

microsoft

-0.72

role

-0.72

owitz

-0.72

replace

-0.70

OLOGY

-0.65

ships

-0.64

×¨

-0.64

ettle

-0.63

endants

-0.63

POSITIVE LOGITS

 dissent

1.26

 impulses

1.00

 dissenting

0.96

 disbelief

0.91

 rebell

0.87

 enthusiasm

0.86

 emotions

0.85

 tendencies

0.84

 creativity

0.83

 runaway

0.83

Activations Density 0.115%

phrases describing attempts to control or restrain something

instances of the word "suppress" and related forms

No Comments

No Known Activations