INDEX

Explanations

instances of words associated with strong negative emotions or reactions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "outrage" and related variations expressing strong emotional responses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

uchin

-0.76

ramer

-0.75

arnaev

-0.71

essions

-0.70

 depress

-0.68

sembly

-0.67

wagen

-0.67

 Solitaire

-0.66

ession

-0.65

stood

-0.65

POSITIVE LOGITS

ously

1.22

 indignation

0.98

iously

0.93

 uproar

0.78

 outrage

0.78

èĥ

0.76

ingly

0.76

EU

0.74

 outcry

0.72

storm

0.72

Activations Density 0.018%

instances of words associated with strong negative emotions or reactions

instances of the word "outrage" and related variations expressing strong emotional responses

No Comments

No Known Activations