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

iland

-0.77

 VIDEOS

-0.75

ãĥ´

-0.73

STAR

-0.72

OUS

-0.69

é¾

-0.69

INGTON

-0.68

OVER

-0.66

GREEN

-0.66

estern

-0.65

POSITIVE LOGITS

who

0.90

who

0.74

 whom

0.70

 whose

0.68

avez

0.66

ept

0.62

 Shinra

0.61

born

0.59

whose

0.59

 discipl

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.