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

©¶æ

-0.70

ream

-0.64

ulkan

-0.64

apers

-0.64

ierre

-0.63

aware

-0.62

edia

-0.61

ocument

-0.61

aper

-0.60

atts

-0.60

POSITIVE LOGITS

 against

1.05

against

0.97

 Against

0.85

 versus

0.77

Against

0.75

rather

0.65

 Opposition

0.64

AGA

0.64

mouth

0.64

ARDIS

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.