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

naissance

-0.76

actionDate

-0.71

ccess

-0.67

lege

-0.66

||||

-0.66

atural

-0.66

ndra

-0.65

wal

-0.65

rency

-0.65

anguage

-0.64

POSITIVE LOGITS

LM

0.78

ICS

0.74

CI

0.66

igham

0.65

DO

0.65

Io

0.64

wav

0.63

NH

0.63

generated

0.63

IU

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.