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

 Citiz

-0.84

 citiz

-0.74

 Forth

-0.73

Vote

-0.71

nt

-0.71

ecided

-0.71

QL

-0.70

vez

-0.70

oun

-0.70

Mus

-0.70

POSITIVE LOGITS

 heights

0.69

 creep

0.69

mph

0.65

 inches

0.64

edience

0.63

 Angels

0.62

 NYPD

0.62

 perk

0.61

 pile

0.61

 sewer

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.