INDEX

Explanations

words or phrases indicating protection, security, or close monitoring

oai_token-act-pair · gpt-3.5-turbo

terms related to security and guarding

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

clerosis

-0.80

adder

-0.72

Kay

-0.68

ODE

-0.67

EVA

-0.66

article

-0.65

thin

-0.64

SQL

-0.63

SW

-0.62

MAP

-0.62

POSITIVE LOGITS

 guarded

1.16

 guarding

1.10

 adolesc

0.81

rets

0.76

acle

0.76

 tradem

0.76

erness

0.76

 secrets

0.75

 guardians

0.73

cheon

0.73

Activations Density 0.020%

words or phrases indicating protection, security, or close monitoring

terms related to security and guarding

No Comments

No Known Activations