INDEX

Explanations

references to specific geographic locations, particularly pertaining to a place called "Oga."

oai_token-act-pair · gpt-3.5-turbo

instances of specific geographical or cultural names

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eways

-0.78

paren

-0.67

 Newsweek

-0.64

advertisement

-0.63

tle

-0.62

infect

-0.62

drug

-0.62

ulent

-0.62

nu

-0.62

onym

-0.61

POSITIVE LOGITS

oga

1.06

 Springs

0.96

issance

0.95

userc

0.85

EStream

0.82

ourt

0.82

 Falls

0.79

pload

0.79

 Valley

0.78

qua

0.78

Activations Density 0.013%

references to specific geographic locations, particularly pertaining to a place called "Oga."

instances of specific geographical or cultural names

No Comments

No Known Activations