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

 queues

-0.74

Ħ¢

-0.73

uffs

-0.72

interpret

-0.68

arettes

-0.68

eston

-0.66

 Interpret

-0.66

 Flow

-0.66

ardon

-0.65

vo

-0.65

POSITIVE LOGITS

bolt

0.73

core

0.65

 Crom

0.62

 Calvin

0.62

born

0.61

''''

0.60

Brook

0.60

 Barney

0.59

eyes

0.59

 Cthulhu

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.