INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

translation

-0.66

Accessory

-0.63

 Treasury

-0.62

intend

-0.62

fu

-0.62

Leader

-0.61

èĢħ

-0.61

mobi

-0.59

][/

-0.59

¶æ

-0.58

POSITIVE LOGITS

ioxide

0.81

 miscar

0.74

je

0.67

aleigh

0.64

},"

0.63

arcity

0.59

 Impossible

0.59

 Earthqu

0.58

ases

0.58

ophy

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.