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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĥĳ

-0.74

 Moons

-0.67

=$

-0.67

âĶ

-0.66

 Fenrir

-0.66

atar

-0.63

hog

-0.60

lam

-0.59

 Roads

-0.59

ãĤ°

-0.59

POSITIVE LOGITS

izont

0.85

essional

0.80

ribe

0.78

ILLE

0.78

heastern

0.72

ilater

0.70

ilage

0.70

knit

0.68

-------

0.67

lex

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.