INDEX

Explanations

phrases related to justifying or criticizing actions

oai_token-act-pair · gpt-3.5-turbo

discussions related to accountability and the justification of actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

prus

-0.79

 Arri

-0.79

thood

-0.75

 Horizons

-0.73

vier

-0.69

phe

-0.69

 Awakens

-0.68

rities

-0.68

usha

-0.68

Luck

-0.66

POSITIVE LOGITS

 unlawful

1.35

 inconsistent

1.32

 contrary

1.27

 unethical

1.24

 improper

1.22

 detrimental

1.21

 inappropriate

1.18

 discriminatory

1.17

 unconstitutional

1.17

repre

1.16

Activations Density 0.186%

phrases related to justifying or criticizing actions

discussions related to accountability and the justification of actions

No Comments

No Known Activations