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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

NX

-0.70

THR

-0.66

 disappoint

-0.66

iets

-0.65

 clitor

-0.65

ega

-0.64

retty

-0.63

iology

-0.62

iates

-0.62

ensis

-0.61

POSITIVE LOGITS

Reloaded

0.75

GROUND

0.73

Untitled

0.72

âĵĺ

0.69

ground

0.66

anton

0.66

 sqor

0.64

inventory

0.63

"},"

0.63

ashington

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.