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

wear

-0.68

region

-0.66

keeper

-0.65

mill

-0.65

fps

-0.64

itt

-0.62

erman

-0.60

CIS

-0.59

ingham

-0.59

yard

-0.59

POSITIVE LOGITS

edIn

0.87

ãĤ¹ãĥĪ

0.86

ADRA

0.78

CLS

0.77

rative

0.75

ubi

0.74

blance

0.73

rul

0.73

URI

0.72

edom

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.