INDEX

Explanations

references to specific names and organizations

oai_token-act-pair · gpt-3.5-turbo

proper nouns and specific references, particularly names and organizations

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

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

acters

-0.77

icious

-0.74

ership

-0.73

CLASSIFIED

-0.72

istic

-0.71

ãĥ³ãĤ¸

-0.70

ãĥģ

-0.70

ãĥ¼ãĥĨãĤ£

-0.69

thouse

-0.68

WAYS

-0.68

POSITIVE LOGITS

bye

0.75

IG

0.75

Dyn

0.74

IE

0.65

lie

0.65

oln

0.65

SPONSORED

0.64

raped

0.64

ellen

0.64

 buckle

0.63

Activations Density 0.022%

references to specific names and organizations

proper nouns and specific references, particularly names and organizations

No Comments

No Known Activations