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

analysis

-0.74

ounds

-0.69

atan

-0.67

ounded

-0.67

leading

-0.67

opot

-0.66

letters

-0.64

iation

-0.64

holder

-0.63

igans

-0.63

POSITIVE LOGITS

senal

0.80

 destro

0.79

ADRA

0.79

 challeng

0.77

 arrang

0.70

 confir

0.69

engeance

0.69

©¶æ¥µ

0.68

ĪĴ

0.65

Eye

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.