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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 suck

-0.73

 scare

-0.71

 wrestle

-0.69

chwitz

-0.66

sembly

-0.65

 staggered

-0.65

 sucked

-0.65

vertisement

-0.65

 tremend

-0.64

terday

-0.64

POSITIVE LOGITS

INST

0.74

Van

0.73

Murray

0.72

ankind

0.71

Mut

0.71

DNA

0.70

idian

0.69

NZ

0.67

ropy

0.67

enum

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.