INDEX

Explanations

proper nouns related to specific places such as colleges and universities

oai_token-act-pair · gpt-3.5-turbo

references to specific names or organizations

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

CPC

-0.69

************

-0.68

perture

-0.67

Favorite

-0.65

 disconnected

-0.63

 Carbuncle

-0.62

 Piercing

-0.61

advertisement

-0.60

UAL

-0.60

 choking

-0.58

POSITIVE LOGITS

 Ober

1.11

lander

1.03

lin

0.99

meyer

0.98

stad

0.95

keley

0.95

liga

0.95

lund

0.94

haus

0.86

itas

0.86

Activations Density 0.021%

proper nouns related to specific places such as colleges and universities

references to specific names or organizations

No Comments

No Known Activations