INDEX

Explanations

explicit statements of clarification or emphasis

oai_token-act-pair · gpt-3.5-turbo

phrases indicating clarity or explicitness in statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 luck

-0.72

ickle

-0.67

Luck

-0.65

luck

-0.65

average

-0.65

 Faul

-0.64

digy

-0.63

 Derby

-0.60

 crane

-0.59

 Luck

-0.58

POSITIVE LOGITS

 displeasure

0.92

 unequivocally

0.87

 emphatically

0.86

 intention

0.85

 disapproval

0.83

 willingness

0.83

 intentions

0.80

ariat

0.80

 commitment

0.79

 disdain

0.79

Activations Density 0.201%

explicit statements of clarification or emphasis

phrases indicating clarity or explicitness in statements

No Comments

No Known Activations