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

ancial

-0.70

ILLE

-0.66

 Accessed

-0.64

 CASE

-0.63

ERAL

-0.62

ittal

-0.61

 VIDE

-0.60

STEM

-0.60

CPC

-0.60

BILL

-0.59

POSITIVE LOGITS

miss

0.77

enge

0.73

ilyn

0.70

Orig

0.69

idepress

0.66

orig

0.64

igma

0.63

gon

0.62

merga

0.62

edin

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.