INDEX

Explanations

cryptic messages encoded in a pattern of symbols

oai_token-act-pair · gpt-3.5-turbo

repeated symbols or punctuations that may indicate emphasis or special formatting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 mathemat

-0.73

 Palestin

-0.70

 controvers

-0.70

 diminishing

-0.70

 incorpor

-0.70

 mosqu

-0.70

 notor

-0.70

 obser

-0.70

 strugg

-0.69

 conclud

-0.68

POSITIVE LOGITS

ï¸ı

0.95

Â¯

0.91

ÃĽ

0.87

Â°

0.85

ï¸

0.82

åĽ

0.82

âĻ

0.80

çľ

0.77

log

0.76

ef

0.76

Activations Density 0.175%

cryptic messages encoded in a pattern of symbols

repeated symbols or punctuations that may indicate emphasis or special formatting

No Comments

No Known Activations