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

chio

-0.82

CG

-0.74

istical

-0.68

ice

-0.68

Record

-0.67

SHIP

-0.67

ongevity

-0.66

amus

-0.66

ealous

-0.66

ential

-0.65

POSITIVE LOGITS

forms

0.67

 shelters

0.65

 subdiv

0.65

Yar

0.63

effic

0.63

 randomized

0.62

omin

0.61

 Archdemon

0.60

 brackets

0.60

 Aval

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.