INDEX

Explanations

phrases related to actions involving unscrewing or unmasking

oai_token-act-pair · gpt-3.5-turbo

terms related to being unscheduled or unclear about certain plans or states

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

Cs

-0.78

 Origins

-0.77

 Strategy

-0.76

pheus

-0.75

 Tags

-0.74

ulhu

-0.73

 Architecture

-0.72

 Barrier

-0.71

OPLE

-0.68

anwhile

-0.68

POSITIVE LOGITS

ishable

1.08

unp

0.94

ashed

0.83

encrypted

0.83

 uncomp

0.83

 concess

0.82

edited

0.82

modified

0.82

leased

0.78

iless

0.78

Activations Density 0.015%

phrases related to actions involving unscrewing or unmasking

terms related to being unscheduled or unclear about certain plans or states

No Comments

No Known Activations