INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

istical

-0.74

 blackmail

-0.72

 postage

-0.69

 gamble

-0.68

 birthday

-0.67

 classification

-0.64

earch

-0.64

 capit

-0.62

 boycott

-0.62

 predicament

-0.62

POSITIVE LOGITS

ĳå£«

0.72

 Camer

0.71

 Grounds

0.70

ocumented

0.67

namese

0.67

lda

0.66

è¦ļéĨĴ

0.65

Loading

0.65

 Marsh

0.65

tta

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.