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

 dysph

-0.77

Ire

-0.72

agascar

-0.72

 susceptible

-0.65

merce

-0.65

 stomach

-0.64

 monarch

-0.64

 conflic

-0.62

ierrez

-0.62

rhy

-0.60

POSITIVE LOGITS

ano

0.85

apo

0.79

ushi

0.77

anan

0.76

esa

0.74

dan

0.71

å¥³

0.69

IDA

0.66

aminer

0.66

aking

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.