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

 recover

-0.71

 cleaner

-0.69

 indexed

-0.64

 inspector

-0.63

å£«

-0.62

 Plato

-0.61

inker

-0.60

 ghosts

-0.59

 ghost

-0.57

 translate

-0.57

POSITIVE LOGITS

ro

1.32

hal

0.81

Ire

0.76

ismo

0.74

ril

0.72

rica

0.71

ting

0.71

ternity

0.71

rio

0.70

chio

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.