INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ndra

-0.66

 Jarrett

-0.65

gray

-0.64

DeV

-0.63

 inertia

-0.61

ŃĶ

-0.61

ABE

-0.61

stories

-0.61

à©

-0.61

ige

-0.60

POSITIVE LOGITS

+#

0.77

ummer

0.73

DragonMagazine

0.72

ãĤ¼

0.65

 CLIENT

0.63

 sublime

0.63

ayson

0.60

fold

0.60

orde

0.59

poon

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.