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

 mathemat

-0.73

 parcels

-0.67

Ń·

-0.65

ĺħ

-0.64

 torches

-0.62

 parcel

-0.62

oes

-0.61

acters

-0.61

 ration

-0.60

 notebooks

-0.60

POSITIVE LOGITS

IUM

0.66

VERTISEMENT

0.63

 RELEASE

0.63

acious

0.62

DJ

0.61

Ã¦

0.59

\"

0.59

\/

0.59

 POLITICO

0.58

ammad

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.