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

 Phant

-0.75

 Nato

-0.71

atars

-0.66

icles

-0.65

 Qatar

-0.65

 Arist

-0.64

clusions

-0.64

Topics

-0.64

plex

-0.63

avi

-0.62

POSITIVE LOGITS

reditary

0.87

CHAT

0.71

anooga

0.69

 buttons

0.65

 cannabin

0.65

pole

0.64

leeve

0.64

emark

0.64

 bookmark

0.63

keeper

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.