INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

arta

-0.85

)).

-0.68

eas

-0.68

]).

-0.65

avorable

-0.62

]),

-0.62

Cu

-0.61

ju

-0.61

saf

-0.60

 Franks

-0.60

POSITIVE LOGITS

Own

0.79

aminer

0.66

laus

0.66

yright

0.65

senal

0.63

oster

0.63

Hay

0.63

inventoryQuantity

0.62

oland

0.62

ventures

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.