INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

patrick

-0.95

rad

-0.77

ãĤ¤ãĥĪ

-0.72

gradation

-0.69

ADA

-0.69

yards

-0.68

kay

-0.68

bett

-0.67

RN

-0.67

 Kits

-0.67

POSITIVE LOGITS

venge

0.84

 overflow

0.67

 Anniversary

0.67

 comma

0.67

uous

0.65

 anniversary

0.64

cs

0.64

gif

0.63

venant

0.63

adelphia

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.