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

ufficient

-0.69

 quarters

-0.68

kun

-0.66

oba

-0.65

ython

-0.65

negie

-0.65

atell

-0.63

arna

-0.63

este

-0.62

 manufact

-0.62

POSITIVE LOGITS

ftime

0.86

++++++++++++++++

0.77

HCR

0.72

_>

0.71

catentry

0.71

 Clip

0.70

Laughs

0.70

 opin

0.68

----------

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.