INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Atk

-0.69

worldly

-0.66

lde

-0.66

udeau

-0.65

yssey

-0.65

rican

-0.65

atform

-0.64

 Tayyip

-0.64

irie

-0.61

wcsstore

-0.61

POSITIVE LOGITS

Item

0.70

bytes

0.69

inge

0.69

igi

0.66

anan

0.66

writ

0.65

aton

0.64

Items

0.64

 transpl

0.64

ghazi

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.