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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Berk

-0.67

pse

-0.66

RS

-0.61

relation

-0.61

feat

-0.60

relations

-0.60

 replication

-0.60

 Nile

-0.59

LCS

-0.59

 diligence

-0.59

POSITIVE LOGITS

 into

1.32

 Into

1.23

 INTO

1.17

into

1.09

 onto

0.91

 away

0.83

 overboard

0.70

imet

0.69

imei

0.67

ãĥ³ãĤ¸

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.