INDEX

Explanations

mentions of ethical or legal misconduct

oai_token-act-pair · gpt-3.5-turbo

instances of the word "wrongdoing" and related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

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

bern

-0.79

population

-0.74

birth

-0.71

etically

-0.70

etic

-0.68

pop

-0.68

etics

-0.65

part

-0.63

berman

-0.62

Dak

-0.62

POSITIVE LOGITS

 wrongdoing

1.33

 unfocusedRange

1.08

 misconduct

0.93

 mishand

0.91

 disclosures

0.89

 Disclosure

0.86

 allegations

0.82

 dealings

0.81

 disclosure

0.79

 abuses

0.77

Activations Density 0.020%

mentions of ethical or legal misconduct

instances of the word "wrongdoing" and related concepts

No Comments

No Known Activations