INDEX

Explanations

the word "Shake"

oai_token-act-pair · gpt-3.5-turbo

occurrences of specific brand names or product-related terms

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

neum

-0.72

 Romania

-0.71

imar

-0.71

 Wales

-0.70

umph

-0.69

 Gorge

-0.69

Scar

-0.67

Ukraine

-0.67

henko

-0.66

hell

-0.66

POSITIVE LOGITS

 Shake

2.17

endo

1.46

 Vive

1.38

 Proposition

1.16

 Loll

1.11

ISA

1.05

 Mecca

1.04

posts

1.02

Yog

0.97

 Kepler

0.97

Activations Density 0.092%

the word "Shake"

occurrences of specific brand names or product-related terms

No Comments

No Known Activations