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

otten

-0.71

furt

-0.67

 Nept

-0.66

Emer

-0.65

 Machina

-0.64

\":

-0.63

 Bened

-0.63

weather

-0.62

EMENT

-0.62

ãĥĳ

-0.62

POSITIVE LOGITS

 mounts

0.69

ords

0.64

 droid

0.64

 Crusher

0.63

 brick

0.63

eah

0.60

arrison

0.60

 coordinate

0.59

 orcs

0.59

 goblin

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.