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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gaard

-1.01

iasm

-0.86

ð

-0.67

¥

-0.64

###

-0.64

idth

-0.64

zhen

-0.63

pak

-0.62

ongyang

-0.61

speak

-0.61

POSITIVE LOGITS

ages

0.65

Rated

0.63

 Mechdragon

0.63

åĤ

0.63

Pry

0.60

ciplinary

0.59

ãĥĩãĤ£

0.59

 Prel

0.59

 Torch

0.58

 substitutes

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.