INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 brut

-0.66

 surprisingly

-0.64

 cous

-0.64

 ballistic

-0.62

nun

-0.62

 conscientious

-0.62

 cave

-0.62

 germ

-0.61

 neighbor

-0.61

 nylon

-0.61

POSITIVE LOGITS

éĹĺ

0.83

uph

0.76

opal

0.74

izons

0.71

Discuss

0.71

 Cosponsors

0.70

*/(

0.70

IFT

0.69

ãģ®å

0.69

lift

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.