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

 Trafford

-0.72

 Gret

-0.67

 Sadd

-0.62

GTA

-0.62

Eng

-0.61

 Gray

-0.61

venge

-0.61

IG

-0.60

 Oval

-0.60

Fal

-0.60

POSITIVE LOGITS

zbollah

0.84

eki

0.73

eks

0.72

odon

0.72

usalem

0.72

wig

0.71

icka

0.71

atic

0.71

hn

0.70

ailand

0.70

Activations Density 0.000%

No Known Activations

This feature has no known activations.