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

olulu

-0.89

arcity

-0.80

ayne

-0.76

nir

-0.73

cano

-0.69

aura

-0.69

ripp

-0.68

arantine

-0.68

 Awakens

-0.66

prep

-0.66

POSITIVE LOGITS

Gay

0.73

ãĤ¨ãĥ«

0.72

itely

0.70

iOS

0.66

ss

0.64

 Chocobo

0.62

CSS

0.62

fox

0.62

IGN

0.61

iffs

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.