INDEX

Explanations

names of government positions followed by the word "secret"

oai_token-act-pair · gpt-3.5-turbo

terms related to confidentiality and secrecy

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Springer

-0.73

 brim

-0.70

à

-0.66

odcast

-0.63

alone

-0.63

asts

-0.61

gaard

-0.60

avers

-0.60

 Surv

-0.59

 Tornado

-0.58

POSITIVE LOGITS

secret

1.35

Secret

1.14

rets

0.92

 secret

0.90

ategic

0.88

terday

0.84

arial

0.84

ulously

0.83

uously

0.78

ories

0.77

Activations Density 0.007%

names of government positions followed by the word "secret"

terms related to confidentiality and secrecy

No Comments

No Known Activations