INDEX

Explanations

words related to shame or shameful actions

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Ajax

-0.85

 Tone

-0.73

 Leth

-0.67

OLOGY

-0.66

 Reach

-0.66

 Luther

-0.64

 unfocusedRange

-0.63

anwhile

-0.62

 ground

-0.61

ãĥ¼ãĥĨãĤ£

-0.60

POSITIVE LOGITS

apesh

1.29

atters

1.29

rapnel

1.24

apeshifter

1.22

oddy

1.20

aded

1.19

aders

1.17

attering

1.15

ackle

1.13

ippers

1.13

Activations Density 0.018%

words related to shame or shameful actions

No Comments

No Known Activations