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

thodox

-0.69

tics

-0.68

conom

-0.67

apon

-0.66

kel

-0.66

Reb

-0.64

 Torah

-0.64

folio

-0.62

olesc

-0.60

utic

-0.58

POSITIVE LOGITS

 Kingdoms

0.66

enne

0.63

"},{"

0.60

ãĥ¼ãĥĨãĤ£

0.59

nu

0.59

Â¶

0.59

 thence

0.59

 Destination

0.58

)+

0.58

ensis

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.