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

piracy

-0.77

glas

-0.70

null

-0.69

 Lizard

-0.68

gra

-0.67

pillar

-0.66

Merit

-0.64

angered

-0.62

inel

-0.61

eryl

-0.61

POSITIVE LOGITS

OUP

0.87

orr

0.70

ATING

0.67

WIND

0.65

 endings

0.63

éŃĶ

0.63

ACK

0.63

CHO

0.63

å£«

0.62

izont

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.