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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ural

-0.75

aneous

-0.69

ermott

-0.66

enary

-0.64

 consent

-0.62

encies

-0.61

 timely

-0.60

 Marketplace

-0.59

 Catalyst

-0.59

 Revel

-0.58

POSITIVE LOGITS

Phase

0.84

mega

0.79

PsyNetMessage

0.76

rawdownloadcloneembedreportprint

0.76

âĹ¼

0.72

Gy

0.71

vid

0.70

Got

0.69

 Toll

0.68

Dim

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.