INDEX

Explanations

mentions of the word "magic."

oai_token-act-pair · gpt-3.5-turbo

references to "magic" or related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 compr

-0.85

 Pearce

-0.67

tz

-0.67

emouth

-0.65

tnc

-0.64

 centr

-0.63

 unmarked

-0.62

çİĭ

-0.62

aque

-0.62

VK

-0.61

POSITIVE LOGITS

Magic

1.25

magic

0.77

catch

0.69

illusion

0.69

copy

0.62

reports

0.61

Focus

0.59

 lash

0.58

 photons

0.58

cars

0.58

Activations Density 0.003%

mentions of the word "magic."

references to "magic" or related concepts

No Comments

No Known Activations