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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ãĤ¦ãĤ¹

-0.85

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.81

 Metatron

-0.77

 lapt

-0.77

 tremend

-0.74

ä½¿

-0.73

otropic

-0.72

 illum

-0.71

ppo

-0.71

BILITIES

-0.71

POSITIVE LOGITS

 Battles

0.87

 Bundy

0.73

awks

0.72

 Payne

0.69

 Playoffs

0.68

iasis

0.68

atoon

0.66

 Battalion

0.66

arrass

0.65

 Ferry

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.