INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

umbn

-0.81

 fortun

-0.73

 Gale

-0.69

 Gutenberg

-0.66

 obscurity

-0.65

 nomine

-0.64

 Philipp

-0.64

 funer

-0.63

 Playboy

-0.63

 Griffith

-0.63

POSITIVE LOGITS

Train

0.86

ÃŁ

0.84

faced

0.76

ski

0.75

per

0.74

rate

0.74

Rate

0.73

orbit

0.73

hyp

0.72

rocket

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.