INDEX

Explanations

words starting with "Se" followed by a high activation number

oai_token-act-pair · gpt-3.5-turbo

occurrences of the character sequence "Se"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Downloadha

-0.76

aneers

-0.74

 wire

-0.73

dayName

-0.69

 hoop

-0.66

 realism

-0.66

CPC

-0.65

 overhead

-0.65

 crib

-0.63

 deed

-0.63

POSITIVE LOGITS

venth

1.15

vere

1.12

quel

1.12

eker

1.09

asons

1.04

ems

1.03

lected

1.03

Se

1.00

ek

0.99

eking

0.99

Activations Density 0.006%

words starting with "Se" followed by a high activation number

occurrences of the character sequence "Se"

No Comments

No Known Activations