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

onal

-0.72

olescent

-0.65

 circuit

-0.64

ourced

-0.63

RC

-0.61

 gasp

-0.59

PIN

-0.59

cial

-0.59

ci

-0.58

 Levine

-0.58

POSITIVE LOGITS

gow

0.84

cov

0.80

 Lann

0.71

 welf

0.67

querque

0.67

 Cheong

0.64

 Evening

0.63

lys

0.63

rys

0.63

 Ezek

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.