INDEX

Explanations

words related to traps, such as "trapdoors" and "lure."

oai_token-act-pair · gpt-3.5-turbo

references to traps and trapping mechanisms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

league

-0.73

edly

-0.69

Merit

-0.67

 normalized

-0.67

sis

-0.66

issance

-0.65

 Kardashian

-0.64

amily

-0.63

 Supplement

-0.63

kees

-0.62

POSITIVE LOGITS

door

1.48

doors

1.33

trap

1.06

finding

0.96

 traps

0.94

 trap

0.91

 Trap

0.86

idon

0.84

ERY

0.79

 Traps

0.79

Activations Density 0.015%

words related to traps, such as "trapdoors" and "lure."

references to traps and trapping mechanisms

No Comments

No Known Activations