INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 aisle

-0.67

pread

-0.65

Sol

-0.64

edin

-0.63

Pad

-0.62

Berry

-0.61

Rah

-0.61

 nomination

-0.60

 solicitation

-0.60

Secret

-0.59

POSITIVE LOGITS

anwhile

0.80

etheless

0.76

 Ogre

0.72

merce

0.71

©¶æ¥µ

0.70

 dstg

0.69

araoh

0.68

externalActionCode

0.68

 fert

0.66

@@

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.