INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 dstg

-0.78

CLOSE

-0.77

IBLE

-0.74

shots

-0.72

dq

-0.72

£ı

-0.67

RAY

-0.66

encies

-0.66

ibilities

-0.64

 inquest

-0.64

POSITIVE LOGITS

ãĥĥãĥĪ

0.84

imar

0.73

vale

0.71

 Wasteland

0.70

enium

0.69

 Diver

0.69

Sense

0.67

epad

0.64

ãĤ¦ãĤ¹

0.63

oS

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.