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

anwhile

-0.83

anga

-0.75

milo

-0.73

geist

-0.73

ued

-0.73

 skelet

-0.71

culosis

-0.68

tumblr

-0.68

interstitial

-0.65

STD

-0.65

POSITIVE LOGITS

mosp

0.94

Buyable

0.70

Corp

0.70

Ps

0.66

Merc

0.63

Pros

0.62

George

0.61

 Modes

0.60

Pocket

0.59

qv

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.