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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ilde

-0.81

UGC

-0.73

laus

-0.72

ilion

-0.71

 Orchestra

-0.70

anguage

-0.67

arest

-0.65

uck

-0.65

orf

-0.65

eele

-0.64

POSITIVE LOGITS

 resil

0.88

 suscept

0.79

 biod

0.78

 Tasman

0.72

vez

0.68

natureconservancy

0.66

 horm

0.66

 behavi

0.62

 amen

0.60

 restores

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.