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

hap

-0.77

usage

-0.76

Downloadha

-0.72

 Anon

-0.70

Wik

-0.70

NPR

-0.68

 Qiao

-0.67

ktop

-0.66

Hy

-0.65

unta

-0.65

POSITIVE LOGITS

opathic

0.76

oS

0.75

versive

0.69

ingly

0.69

croft

0.69

otropic

0.68

 gravy

0.67

lette

0.67

osure

0.66

ently

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.