INDEX

Explanations

empty or unused sections in the text

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

owa

-0.90

aston

-0.82

soever

-0.82

ansk

-0.81

eto

-0.75

HCR

-0.73

oho

-0.72

leigh

-0.72

oso

-0.70

osh

-0.68

POSITIVE LOGITS

itant

0.69

Acknowled

0.68

ministic

0.68

 Sources

0.68

aneous

0.65

 Handling

0.63

è¯

0.62

VERTISEMENT

0.59

 Alchemist

0.59

 Origin

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.