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

emouth

-0.70

nels

-0.68

lers

-0.66

uits

-0.66

inyl

-0.66

umpy

-0.66

secut

-0.64

idents

-0.63

 Canaver

-0.63

 Flavoring

-0.63

POSITIVE LOGITS

clus

0.69

aire

0.68

 Service

0.67

 ASUS

0.65

AX

0.64

 Achilles

0.63

asio

0.62

enge

0.61

 Turing

0.61

Eid

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.