INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 comrade

-0.70

èĢħ

-0.66

Jr

-0.61

itton

-0.61

 curator

-0.61

senal

-0.59

 arte

-0.59

 Proud

-0.58

 Johannes

-0.58

Ki

-0.57

POSITIVE LOGITS

illance

0.71

NF

0.66

degree

0.65

achusetts

0.65

rared

0.65

zona

0.63

intent

0.61

rophe

0.59

esame

0.59

ivia

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.