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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 minds

-0.73

 laundering

-0.72

~~~~

-0.66

 sustained

-0.65

 clearing

-0.65

å°Ĩ

-0.64

ded

-0.62

 wrath

-0.62

 reper

-0.61

è¯

-0.60

POSITIVE LOGITS

igmat

0.88

 Drift

0.86

alog

0.78

iHUD

0.77

ategory

0.75

endiary

0.72

ikarp

0.71

risome

0.71

atell

0.71

itus

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.