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

fet

-0.77

electric

-0.70

usp

-0.69

 indexed

-0.68

fixed

-0.66

perty

-0.65

acid

-0.65

 arous

-0.65

ramid

-0.65

SELECT

-0.65

POSITIVE LOGITS

 Meadow

0.77

stal

0.68

 alike

0.65

 Skies

0.65

 Ezek

0.64

esan

0.64

bucks

0.64

gra

0.64

Kry

0.63

¶ħ

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.