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

assadors

-0.65

yip

-0.64

sonian

-0.63

 totaling

-0.60

emouth

-0.60

 portfolio

-0.59

ÃŃs

-0.58

viz

-0.58

 showcased

-0.57

 trend

-0.57

POSITIVE LOGITS

 è£ıè

1.14

ãĥ«

0.71

ãĤ´

0.70

 Dise

0.69

 Bundy

0.69

 Rape

0.68

rag

0.68

ãĤ´ãĥ³

0.68

ãĥ¼ãĥĨ

0.68

cised

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.