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

rolet

-0.69

TMZ

-0.64

lesi

-0.63

tarians

-0.62

ktop

-0.61

alg

-0.60

CNN

-0.60

Ake

-0.60

ationally

-0.59

arling

-0.59

POSITIVE LOGITS

Ire

0.64

 seiz

0.64

senal

0.64

ieu

0.63

 Whilst

0.63

pak

0.61

scribe

0.61

PF

0.60

yz

0.59

PG

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.