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

ya

-0.80

schild

-0.70

kefeller

-0.68

Circ

-0.67

atche

-0.65

 decay

-0.64

 nasal

-0.64

 shells

-0.61

egal

-0.61

chev

-0.60

POSITIVE LOGITS

esta

0.76

 Guantanamo

0.68

 Assassin

0.67

VP

0.66

ãģ®ç

0.62

\\\\\\\\\\\\\\\\

0.60

fight

0.60

endered

0.60

 Tinker

0.59

 FOIA

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.