INDEX

Explanations

words related to safety or security

oai_token-act-pair · gpt-3.5-turbo

references to safety and risk in contexts related to technology or environment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Dough

-0.84

Beer

-0.80

 Hamilton

-0.73

 Kling

-0.70

beer

-0.69

DM

-0.68

reference

-0.67

Marx

-0.67

unit

-0.67

argon

-0.66

POSITIVE LOGITS

saf

2.91

 Safari

2.32

saf

1.57

 Sands

1.46

 Yosemite

1.20

 Bots

1.20

 Watkins

1.19

 Sere

1.15

Maw

1.12

 pard

1.08

Activations Density 0.049%

words related to safety or security

references to safety and risk in contexts related to technology or environment

No Comments

No Known Activations