INDEX

Explanations

statements related to strong negative emotions, particularly anger

oai_token-act-pair · gpt-3.5-turbo

expressions of anger

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

IFrame

Link

•layer 3 - anger

No Comments

Negative Logits

artifacts

-0.70

 livest

-0.70

 Places

-0.69

arent

-0.68

ums

-0.67

glas

-0.67

issue

-0.67

agate

-0.66

sites

-0.66

 Vide

-0.65

POSITIVE LOGITS

 anger

0.91

ingly

0.89

fulness

0.87

 indignation

0.86

ful

0.78

 resentment

0.77

 fury

0.72

 rage

0.71

 outburst

0.69

wart

0.69

Activations Density 0.013%

statements related to strong negative emotions, particularly anger

expressions of anger

No Comments

No Known Activations