INDEX

Explanations

sequences of characters that likely indicate a formatting issue or error

oai_token-act-pair · gpt-3.5-turbo

sequences of repeated characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Peb

-0.73

arlane

-0.73

 strap

-0.73

oller

-0.70

 orbit

-0.68

 portable

-0.68

olor

-0.65

ducers

-0.65

 duct

-0.63

agers

-0.63

POSITIVE LOGITS

——

1.44

————

1.40

—-

1.28

————————

1.21

————————————————

1.19

––

1.07

DragonMagazine

1.03

Lenin

1.01

DOWN

1.01

âĢ¦âĢ¦

0.97

Activations Density 0.012%

sequences of characters that likely indicate a formatting issue or error

sequences of repeated characters

No Comments

No Known Activations