INDEX

Explanations

words and phrases related to evaluating choices and decisions

oai_token-act-pair · gpt-3.5-turbo

expressions related to moral and ethical judgments

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

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

çīĪ

-0.74

anwhile

-0.66

cellaneous

-0.63

enture

-0.62

swick

-0.62

adel

-0.58

mentioned

-0.58

ierrez

-0.57

elve

-0.56

»Ĵ

-0.56

POSITIVE LOGITS

 worthwhile

0.71

acea

0.67

opathy

0.66

 unconstitutional

0.66

 somehow

0.66

 pedoph

0.66

 worth

0.63

 imminent

0.63

 acceptable

0.63

 beneficial

0.61

Activations Density 0.989%

words and phrases related to evaluating choices and decisions

expressions related to moral and ethical judgments

No Comments

No Known Activations