INDEX

Explanations

hashtags

oai_token-act-pair · gpt-3.5-turbo

hashtags, indicating social media engagement and topics of discussion

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

 cler

-0.82

 slips

-0.80

 tabl

-0.74

 Ambrose

-0.73

 Gors

-0.72

 tongues

-0.72

oun

-0.71

 commission

-0.69

 shuff

-0.69

 clocks

-0.69

POSITIVE LOGITS

Gamer

1.31

RIP

1.22

Ask

1.21

define

1.15

cdn

1.10

DIV

1.09

Never

1.08

gamer

1.07

Fake

1.05

MAG

1.05

Activations Density 0.026%

hashtags

hashtags, indicating social media engagement and topics of discussion

No Comments

No Known Activations