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

ablish

-0.86

venture

-0.82

wal

-0.77

POS

-0.70

itect

-0.68

erman

-0.68

ablishment

-0.68

istan

-0.67

travel

-0.66

rin

-0.65

POSITIVE LOGITS

 Skydragon

0.70

 juven

0.69

 ensu

0.63

 sear

0.62

 distinctive

0.62

 destro

0.62

 submar

0.61

 welf

0.61

 weakest

0.60

 calves

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.