INDEX

Explanations

proper nouns and brand names with varying degrees of activation strength

oai_token-act-pair · gpt-3.5-turbo

abbreviations and acronyms related to companies or brands

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Takeru

-0.65

 Creed

-0.63

 Betsy

-0.63

Zac

-0.59

ullah

-0.58

bender

-0.57

tc

-0.57

 commissions

-0.57

Suz

-0.56

swick

-0.56

POSITIVE LOGITS

bilt

1.14

namese

0.88

amins

0.87

emort

0.86

 versa

0.84

²¾

0.84

iane

0.79

oslav

0.78

Leaks

0.76

ilated

0.76

Activations Density 0.118%

proper nouns and brand names with varying degrees of activation strength

abbreviations and acronyms related to companies or brands

No Comments

No Known Activations