INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

lopp

-0.89

APD

-0.75

ÃŁ

-0.74

 CODE

-0.71

division

-0.70

wered

-0.70

rule

-0.69

Dispatch

-0.68

ħĭ

-0.67

 Dise

-0.66

POSITIVE LOGITS

 immortal

0.72

 enough

0.66

OTOS

0.64

 sufficient

0.60

Alt

0.60

 circ

0.59

 freezes

0.58

 meas

0.57

 sufficiently

0.57

 Hats

0.56

Activations Density 0.000%

No Known Activations

This feature has no known activations.