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

 Passage

-0.76

 Hungry

-0.76

 Stab

-0.73

 Hamm

-0.69

 formation

-0.68

Dj

-0.68

 openings

-0.66

 Weir

-0.66

Chr

-0.65

 Judd

-0.64

POSITIVE LOGITS

aii

0.84

umbo

0.81

ICAN

0.77

UP

0.76

aa

0.76

exclusive

0.76

OPLE

0.75

Pear

0.75

emort

0.74

ffic

0.74

Activations Density 0.000%

No Known Activations

This feature has no known activations.