INDEX

Explanations

phrases related to accusations or allegations of misconduct or wrongdoing

oai_token-act-pair · gpt-3.5-turbo

allegations and accusations of misconduct or illegal activities

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 partName

-0.83

enment

-0.81

ciating

-0.75

Tokens

-0.75

Score

-0.75

gence

-0.73

Zone

-0.70

english

-0.69

Wem

-0.69

apt

-0.69

POSITIVE LOGITS

 improperly

1.17

 mishand

1.16

 inappropriately

1.15

 unlawfully

1.14

 misled

1.10

 misconduct

1.05

 improper

1.05

 falsely

1.05

 plagiar

1.04

 wiret

1.04

Activations Density 0.468%

phrases related to accusations or allegations of misconduct or wrongdoing

allegations and accusations of misconduct or illegal activities

No Comments

No Known Activations