INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

enza

-0.94

apa

-0.78

qua

-0.69

¶æ

-0.68

alde

-0.68

gam

-0.67

egu

-0.67

bilt

-0.67

gments

-0.66

alach

-0.65

POSITIVE LOGITS

 Bots

0.77

 probes

0.68

 Grimm

0.65

 Trojan

0.61

 intrigue

0.60

 intrusive

0.59

oval

0.58

 Babel

0.58

 Blair

0.58

 probe

0.57

Activations Density 0.000%

No Known Activations

This feature has no known activations.