INDEX

Explanations

words related to negative behavior or actions, specifically focusing on harassment

oai_token-act-pair · gpt-3.5-turbo

instances of the word "harassment" in various contexts

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

rians

-0.76

ACTED

-0.76

archs

-0.71

essential

-0.71

éĹĺ

-0.71

obb

-0.71

arch

-0.70

rich

-0.69

ramid

-0.68

stanbul

-0.67

POSITIVE LOGITS

 harass

1.07

 harassment

1.02

 harassing

0.92

 harassed

0.91

 stalking

0.84

 accus

0.78

assment

0.78

 tactics

0.73

lords

0.73

 complaints

0.72

Activations Density 0.017%

words related to negative behavior or actions, specifically focusing on harassment

instances of the word "harassment" in various contexts

No Comments

No Known Activations