INDEX

Explanations

words related to negative aspects, particularly flaws or vulnerabilities

oai_token-act-pair · gpt-3.5-turbo

terms related to flaws and imperfections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Carbuncle

-0.75

NAS

-0.72

scl

-0.68

 Jackets

-0.67

NetMessage

-0.66

ahon

-0.65

rontal

-0.65

vag

-0.64

arson

-0.62

 Archangel

-0.62

POSITIVE LOGITS

lessly

1.20

 flaw

0.98

 flaws

0.97

acies

0.90

nah

0.83

lessness

0.79

ulence

0.79

 plag

0.78

nesses

0.74

 compromises

0.73

Activations Density 0.016%

words related to negative aspects, particularly flaws or vulnerabilities

terms related to flaws and imperfections

No Comments

No Known Activations