INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

eki

-0.78

jri

-0.72

tnc

-0.72

 constitu

-0.70

uesday

-0.70

alan

-0.69

iru

-0.66

ailable

-0.66

agara

-0.66

Cha

-0.66

POSITIVE LOGITS

 Haitian

0.67

acted

0.66

throp

0.62

 ants

0.59

 sells

0.59

insky

0.57

 traveled

0.57

 sold

0.57

 unpop

0.56

 livest

0.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.