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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rael

-0.79

 Asgard

-0.76

 Dres

-0.71

 respectively

-0.71

 Antiqu

-0.70

[[

-0.69

 fundament

-0.67

Lep

-0.64

 Atlantis

-0.64

 Skydragon

-0.63

POSITIVE LOGITS

Sales

0.70

ially

0.67

UGH

0.67

Delivery

0.66

0.65

0.64

trump

0.63

scan

0.63

UST

0.62

ovember

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.