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

ufact

-0.74

gui

-0.74

 indisp

-0.71

 overhe

-0.70

vel

-0.67

 endors

-0.64

ributes

-0.61

 cheaply

-0.61

EPA

-0.61

eco

-0.60

POSITIVE LOGITS

 Morty

0.80

isSpecialOrderable

0.74

tein

0.73

argon

0.68

 Beir

0.67

lock

0.67

ãĥ´ãĤ¡

0.65

ajor

0.62

 Grimoire

0.62

 Ratio

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.