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

 tremend

-0.77

 guiActiveUn

-0.76

 brut

-0.71

ule

-0.67

 garg

-0.66

è¦ļéĨĴ

-0.64

å¥

-0.64

 creations

-0.64

 fortunes

-0.63

 explor

-0.62

POSITIVE LOGITS

 Funds

0.78

edu

0.72

 Coins

0.71

 actionGroup

0.67

amazon

0.66

 Mayweather

0.66

keley

0.66

ota

0.65

 Advance

0.65

sburgh

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.