INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 WATCHED

-0.76

 Takeru

-0.75

vernment

-0.75

gart

-0.70

netflix

-0.70

 Surv

-0.70

 Niet

-0.68

 reluct

-0.68

airo

-0.66

 penetrate

-0.66

POSITIVE LOGITS

TPS

0.62

arrett

0.62

coat

0.61

CCC

0.60

 alarm

0.60

è£

0.60

cas

0.59

iesel

0.59

Nun

0.58

paren

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.