INDEX

Explanations

symbols or patterns repeated in a structured manner

oai_token-act-pair · gpt-3.5-turbo

special characters and symbols

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

 manif

-0.78

 GOODMAN

-0.74

 incorpor

-0.73

 pockets

-0.70

çĶ°

-0.70

 mathemat

-0.70

 Chambers

-0.69

 fabrication

-0.69

 fortun

-0.69

 confir

-0.69

POSITIVE LOGITS

amp

1.49

gt

1.26

lt

1.03

album

0.99

amps

0.99

ohm

0.94

display

0.94

amping

0.93

anger

0.90

tex

0.90

Activations Density 0.008%

symbols or patterns repeated in a structured manner

special characters and symbols

No Comments

No Known Activations