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

hops

-0.69

andel

-0.69

ItemImage

-0.66

aughtered

-0.66

 acron

-0.64

alled

-0.63

ossible

-0.62

rote

-0.62

etheless

-0.61

trop

-0.61

POSITIVE LOGITS

 Guth

0.82

 Nieto

0.73

Neh

0.73

Fn

0.70

 Deity

0.67

 Crush

0.66

Smy

0.65

 Bacon

0.65

epad

0.64

ixon

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.