INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Coffin

-0.71

ĸļ

-0.71

obs

-0.67

ucker

-0.64

ĪĴ

-0.64

ucks

-0.62

icles

-0.62

omy

-0.61

anas

-0.61

usa

-0.60

POSITIVE LOGITS

 amongst

1.22

 among

1.21

among

1.07

Among

0.79

ktop

0.75

 Among

0.75

 Palest

0.73

 encount

0.71

 vulner

0.71

PE

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.