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

oka

-0.84

tein

-0.79

 ILCS

-0.78

eny

-0.72

mson

-0.69

ean

-0.67

drivers

-0.66

ayne

-0.66

uku

-0.66

ibrary

-0.65

POSITIVE LOGITS

iP

0.82

 Router

0.64

senal

0.64

\",

0.61

FP

0.60

lio

0.59

Inside

0.59

 Meteor

0.58

iT

0.58

Ans

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.