INDEX

Explanations

numerical references, specifically ones indicating an upward trend or amount

oai_token-act-pair · gpt-3.5-turbo

phrases involving the concept of "upwards" or increasing values

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tein

-0.73

 Secrets

-0.71

 Doctor

-0.71

 Lieberman

-0.71

 Mamm

-0.70

 Thing

-0.69

Reviewed

-0.69

nen

-0.68

Doctor

-0.68

zig

-0.67

POSITIVE LOGITS

¥ŀ

1.24

etheless

1.08

srf

0.97

 mathemat

0.96

©¶æ¥µ

0.95

 proport

0.93

 earthqu

0.93

 tremend

0.93

 upward

0.91

 ende

0.91

Activations Density 0.003%

numerical references, specifically ones indicating an upward trend or amount

phrases involving the concept of "upwards" or increasing values

No Comments

No Known Activations