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

 pots

-0.65

 Bought

-0.64

stocks

-0.64

ooks

-0.62

 lumber

-0.61

 Bowman

-0.59

 baskets

-0.59

rower

-0.59

itiveness

-0.58

HL

-0.58

POSITIVE LOGITS

vre

0.78

izoph

0.75

cember

0.73

 Mastery

0.70

REDACTED

0.67

Reloaded

0.67

selves

0.66

 Rune

0.66

 Osiris

0.65

ADRA

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.