INDEX

Explanations

percentages or numerical figures in texts

oai_token-act-pair · gpt-3.5-turbo

grammatical structures indicating relationships or connections among subjects and actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ufact

-0.67

Lear

-0.66

EngineDebug

-0.64

ONSORED

-0.62

HUD

-0.62

Angelo

-0.61

Bey

-0.60

ãĥĥãĥī

-0.59

Cop

-0.59

boarding

-0.59

POSITIVE LOGITS

 comprise

1.32

are

1.30

 constitute

1.27

 embody

1.23

 collide

1.22

 represent

1.21

 await

1.17

 reside

1.17

 coincide

1.16

 resemble

1.16

Activations Density 0.423%

percentages or numerical figures in texts

grammatical structures indicating relationships or connections among subjects and actions

No Comments

No Known Activations