INDEX

Explanations

phrases indicating perception or evaluation of a situation

oai_token-act-pair · gpt-3.5-turbo

phrases that relate to perceived value or worth

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

milo

-0.81

anooga

-0.74

ARS

-0.73

bral

-0.72

 ashore

-0.72

abases

-0.70

ests

-0.70

"]=>

-0.68

enberg

-0.68

ä»£

-0.67

POSITIVE LOGITS

 contradiction

0.88

 innocuous

0.87

 quaint

0.86

 inev

0.83

 kindred

0.78

ut

0.74

oxy

0.74

 impossibility

0.72

 coincidence

0.69

 utopian

0.69

Activations Density 0.184%

phrases indicating perception or evaluation of a situation

phrases that relate to perceived value or worth

No Comments

No Known Activations