INDEX

Explanations

mentions of losing or loss in a competitive context

oai_token-act-pair · gpt-3.5-turbo

terms associated with defeat or negative outcomes in sports contexts

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

tan

-0.85

enegger

-0.80

ridges

-0.77

azon

-0.73

oola

-0.73

onomic

-0.70

mun

-0.68

lon

-0.67

height

-0.66

ature

-0.66

POSITIVE LOGITS

 streak

0.85

 foes

0.81

 streaks

0.81

 miser

0.75

esses

0.72

 aversion

0.69

 opponents

0.67

 horribly

0.67

 boss

0.67

bite

0.66

Activations Density 0.039%

mentions of losing or loss in a competitive context

terms associated with defeat or negative outcomes in sports contexts

No Comments

No Known Activations