INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

forward

-0.74

han

-0.70

 forward

-0.67

inventoryQuantity

-0.62

 disclosed

-0.62

 haste

-0.61

 bore

-0.60

Ã¡

-0.60

 harbor

-0.60

 ener

-0.60

POSITIVE LOGITS

Adapt

0.75

Redditor

0.73

 Tend

0.71

idas

0.69

Dual

0.68

 Percent

0.67

devices

0.67

rite

0.65

 Wrap

0.65

 Medals

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.