INDEX

Explanations

No Explanations Found

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 appreci

-0.70

 mascul

-0.69

kees

-0.68

inyl

-0.68

anka

-0.66

idon

-0.65

iam

-0.65

 Hait

-0.65

 dors

-0.63

 desper

-0.62

POSITIVE LOGITS

bsp

0.84

iliate

0.80

multi

0.71

nesday

0.70

np

0.68

CoC

0.66

MU

0.65

 merge

0.64

ovie

0.63

 Begins

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.