INDEX

Explanations

mentions of skills and abilities

oai_token-act-pair · gpt-3.5-turbo

references to various skill levels or mentions of skills in a context that suggests proficiency

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 EDITION

-0.75

 Corinthians

-0.70

 Reloaded

-0.69

 Amen

-0.67

 Cosby

-0.66

 Manz

-0.66

clipse

-0.65

Wem

-0.64

 Orwell

-0.63

 Express

-0.63

POSITIVE LOGITS

fully

1.10

ful

1.03

poke

0.97

anooga

0.96

smanship

0.91

cius

0.86

skill

0.84

chery

0.84

fighter

0.83

 skills

0.82

Activations Density 0.014%

mentions of skills and abilities

references to various skill levels or mentions of skills in a context that suggests proficiency

No Comments

No Known Activations