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

ims

-0.75

arez

-0.71

rez

-0.69

rencies

-0.68

 natives

-0.67

 Emin

-0.66

iors

-0.66

»Ĵ

-0.65

Imm

-0.65

ifix

-0.65

POSITIVE LOGITS

flake

0.69

dit

0.68

mable

0.68

 Hilbert

0.64

kson

0.63

Kod

0.62

 autop

0.61

 knots

0.60

 Hemp

0.59

ragon

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.