INDEX

Explanations

references to facial features such as expressions, bone structure, and injuries

oai_token-act-pair · gpt-3.5-turbo

references to facial features and injuries

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ensable

-0.85

hedon

-0.77

aday

-0.70

 Clever

-0.70

geist

-0.68

uden

-0.67

arity

-0.66

 Pharma

-0.66

acent

-0.65

 Emerson

-0.65

POSITIVE LOGITS

 hair

1.00

 expressions

0.98

 facial

0.96

 recognition

0.91

 beard

0.88

 tattoos

0.86

 Hair

0.86

 portraits

0.85

 impressions

0.85

 expression

0.84

Activations Density 0.018%

references to facial features such as expressions, bone structure, and injuries

references to facial features and injuries

No Comments

No Known Activations