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

isite

-0.68

velength

-0.68

psons

-0.67

EStreamFrame

-0.67

ensen

-0.65

tu

-0.65

 specialize

-0.64

cible

-0.60

licks

-0.58

mberg

-0.58

POSITIVE LOGITS

inventoryQuantity

0.74

NESS

0.73

Sov

0.72

 Idea

0.70

romeda

0.68

 advoc

0.68

ASY

0.68

politics

0.67

 Tuls

0.67

selves

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.