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

hots

-0.73

etimes

-0.71

phies

-0.67

 dumps

-0.65

 Dresden

-0.64

gars

-0.63

aches

-0.62

 highs

-0.62

cu

-0.60

 charisma

-0.60

POSITIVE LOGITS

é¾įåĸļå£«

0.85

Reviewer

0.81

Reviewed

0.80

Route

0.69

Dispatch

0.68

åĽ

0.66

 Review

0.66

 Altern

0.65

 Susp

0.65

 Runner

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.