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

 Alley

-0.73

etus

-0.72

fly

-0.72

minster

-0.70

IUM

-0.68

GW

-0.66

iculture

-0.66

ania

-0.66

pora

-0.65

lem

-0.65

POSITIVE LOGITS

credit

0.70

Crunch

0.70

#$

0.65

 Canaver

0.65

ospels

0.63

question

0.62

ONSORED

0.62

company

0.61

 Sachs

0.60

 SHOULD

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.