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

Reviewer

-0.78

 Rebell

-0.73

onew

-0.70

 Enlight

-0.69

ilogy

-0.66

DEF

-0.65

wei

-0.64

 Elys

-0.64

aith

-0.64

wic

-0.63

POSITIVE LOGITS

uchi

0.73

 notation

0.72

advertisement

0.71

ription

0.70

definition

0.68

ibles

0.68

termination

0.66

tera

0.65

detail

0.64

entry

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.