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

 Fitness

-0.78

 Personally

-0.69

 Improvement

-0.68

Wheel

-0.66

 Predict

-0.65

 Respons

-0.64

 Requires

-0.64

 Tactics

-0.64

hazard

-0.63

 Recreation

-0.63

POSITIVE LOGITS

oyd

0.80

asta

0.75

dos

0.75

agram

0.75

anism

0.74

oming

0.71

ajor

0.70

grave

0.66

eport

0.66

%%%%

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.