INDEX

Explanations

statements or situations that involve threats or danger

oai_token-act-pair · gpt-3.5-turbo

instances of the word "threatening."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

mys

-1.12

arist

-0.91

algia

-0.88

bits

-0.82

ellect

-0.82

tan

-0.80

ria

-0.79

otide

-0.76

urgy

-0.76

gart

-0.76

POSITIVE LOGITS

 threatening

1.17

 threats

1.13

 menacing

1.07

 threaten

1.06

 undermin

1.05

 challeng

1.00

 threatened

0.94

 endanger

0.94

 threatens

0.92

 proble

0.91

Activations Density 0.012%

statements or situations that involve threats or danger

instances of the word "threatening."

No Comments

No Known Activations