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

ãĥ´ãĤ¡

-0.68

ingham

-0.66

RM

-0.66

Ott

-0.65

neb

-0.62

Community

-0.62

Liter

-0.61

ById

-0.61

 Wyatt

-0.61

æľ

-0.59

POSITIVE LOGITS

ndra

0.85

brates

0.77

arc

0.69

 quake

0.67

natureconservancy

0.63

 blessed

0.63

bidden

0.62

 Rage

0.62

 rendition

0.61

 banished

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.