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

ummies

-0.76

utterstock

-0.72

]).

-0.68

vation

-0.67

phis

-0.66

ulatory

-0.65

clave

-0.65

chest

-0.65

llo

-0.65

]),

-0.64

POSITIVE LOGITS

bec

0.69

Reviewer

0.69

 galactic

0.64

 artif

0.63

 interstellar

0.61

 Crew

0.61

ELY

0.61

 Skywalker

0.60

 assistant

0.59

 chancellor

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.