INDEX

Explanations

acronyms related to organizations or groups

oai_token-act-pair · gpt-3.5-turbo

references to the term "PA" (likely an abbreviation or acronym)

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cius

-0.79

 Norse

-0.67

 Cyborg

-0.65

 ethn

-0.64

 multicultural

-0.62

 enchanted

-0.60

Boh

-0.60

 Aether

-0.60

 determining

-0.60

icators

-0.60

POSITIVE LOGITS

PA

4.24

PA

1.98

SPA

1.68

pa

1.65

MI

1.41

VA

1.40

PLA

1.40

Pa

1.39

RA

1.37

PI

1.37

Activations Density 0.010%

acronyms related to organizations or groups

references to the term "PA" (likely an abbreviation or acronym)

No Comments

No Known Activations