INDEX

Explanations

verbs or phrases indicating information disclosure or lack thereof

oai_token-act-pair · gpt-3.5-turbo

statements related to non-disclosure and specification of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

joice

-0.63

ngth

-0.61

jam

-0.60

ruction

-0.60

oir

-0.60

Ã©t

-0.59

cosystem

-0.58

ãĤ¨

-0.56

CHA

-0.56

 âī

-0.54

POSITIVE LOGITS

 specifics

1.33

 whether

1.29

nor

1.03

whether

1.02

 specific

1.00

why

0.99

 particulars

0.99

any

0.92

how

0.91

 exact

0.89

Activations Density 0.142%

verbs or phrases indicating information disclosure or lack thereof

statements related to non-disclosure and specification of information

No Comments

No Known Activations