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

 individually

-0.68

enburg

-0.65

eters

-0.63

Ger

-0.62

 Plex

-0.61

brance

-0.61

attr

-0.60

 Appalach

-0.60

unte

-0.60

 ultras

-0.59

POSITIVE LOGITS

0.74

awa

0.68

 RELEASE

0.64

 commentary

0.63

Vice

0.63

illin

0.63

Cy

0.62

Delivery

0.61

alys

0.61

BLIC

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.