INDEX

Explanations

words related to various types of allegations and investigations

oai_token-act-pair · gpt-3.5-turbo

phrases related to allegations 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 @ 11-res-jb

Configuration

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

rieve

-0.76

answer

-0.71

 Unsure

-0.69

ciating

-0.69

 partName

-0.67

reci

-0.67

dt

-0.66

DragonMagazine

-0.66

onds

-0.66

 ç¥ŀ

-0.66

POSITIVE LOGITS

 wrongdoing

1.61

 misconduct

1.58

 improper

1.38

 corruption

1.37

 unethical

1.32

 malf

1.32

 irregularities

1.30

 incompetence

1.28

 inappropriate

1.28

 bias

1.27

Activations Density 0.270%

words related to various types of allegations and investigations

phrases related to allegations of misconduct or illegal activities

No Comments

No Known Activations