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

uctions

-0.75

myra

-0.67

gaard

-0.67

iator

-0.66

ãĥĥãĥī

-0.66

ovember

-0.64

iability

-0.63

CVE

-0.62

iating

-0.62

iable

-0.62

POSITIVE LOGITS

olla

0.72

 Lean

0.62

rise

0.62

scape

0.62

eous

0.61

ship

0.60

smart

0.58

Dub

0.58

sole

0.58

micro

0.57

Activations Density 0.000%

No Known Activations

This feature has no known activations.