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

tten

-0.78

ICES

-0.75

orses

-0.75

orse

-0.68

ð

-0.66

ikk

-0.64

ajo

-0.64

 coales

-0.63

aken

-0.63

atem

-0.63

POSITIVE LOGITS

taboola

0.89

 Queensland

0.75

itudinal

0.73

 Rampage

0.72

 Curve

0.68

pec

0.68

TTL

0.66

 trivia

0.66

 Revenge

0.65

 Zoom

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.