INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

orer

-0.78

ents

-0.75

ault

-0.70

edom

-0.67

oris

-0.67

 faults

-0.67

orem

-0.66

 triggered

-0.62

orers

-0.60

Hidden

-0.60

POSITIVE LOGITS

thought

0.76

hire

0.68

swer

0.66

phe

0.62

noon

0.62

roc

0.62

 Kushner

0.61

 Shia

0.61

Ëľ

0.61

sha

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.