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

emale

-0.82

hemy

-0.77

ktop

-0.76

itism

-0.73

ridor

-0.69

acebook

-0.69

essage

-0.69

ylum

-0.69

IER

-0.68

NetMessage

-0.68

POSITIVE LOGITS

enne

0.67

 vantage

0.67

 Direction

0.65

ptr

0.63

atts

0.62

 Views

0.61

(),

0.60

 backdrop

0.58

æ©Ł

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.