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

abase

-0.94

anmar

-0.90

zbollah

-0.80

andise

-0.77

agascar

-0.75

\\\\\\\\

-0.73

igl

-0.73

zn

-0.73

phabet

-0.72

orgetown

-0.71

POSITIVE LOGITS

MSG

0.61

0.60

 platforms

0.60

 capsules

0.60

ages

0.57

Ð¼

0.57

__

0.56

'.

0.55

age

0.53

CCP

0.53

Activations Density 0.000%

No Known Activations

This feature has no known activations.