INDEX

Explanations

derogatory comments and remarks made online

oai_token-act-pair · gpt-3.5-turbo

instances of derogatory or insulting comments and expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 rebuild

-0.82

verning

-0.81

 rebuilding

-0.78

 glim

-0.77

 rebuilt

-0.74

 taxed

-0.72

 stabilize

-0.72

quartered

-0.71

ensable

-0.69

 workings

-0.68

POSITIVE LOGITS

Redditor

1.15

Reddit

1.04

 retweet

1.03

tumblr

0.98

TMZ

0.97

 tweets

0.96

 Tumblr

0.95

 selfies

0.95

 derogatory

0.93

 misogyn

0.93

Activations Density 0.647%

derogatory comments and remarks made online

instances of derogatory or insulting comments and expressions

No Comments

No Known Activations