INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

grass

-0.76

iba

-0.76

hof

-0.71

ĸļ

-0.70

rye

-0.68

THC

-0.67

season

-0.66

imaru

-0.66

oven

-0.65

winner

-0.65

POSITIVE LOGITS

 Cancel

0.85

 property

0.70

 Strikes

0.65

yrights

0.65

ITED

0.62

 rhetorical

0.61

 expiration

0.60

 Cance

0.60

 istg

0.60

 ambiguous

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.