INDEX

Explanations

phrases indicating evaluation or comparison of situations

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

words and phrases related to moral comparisons and ethical dilemmas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ĸļ

-0.75

clad

-0.68

execute

-0.67

æ©

-0.66

STER

-0.65

PsyNetMessage

-0.64

Jet

-0.64

factor

-0.64

 Borough

-0.63

POSITIVE LOGITS

 evil

1.10

bad

1.06

 evils

0.93

ill

0.93

 detrim

0.91

evil

0.90

 downside

0.90

 worst

0.90

 Evil

0.89

 harms

0.89

Activations Density 0.156%

phrases indicating evaluation or comparison of situations

words and phrases related to moral comparisons and ethical dilemmas

No Comments

No Known Activations