INDEX

Explanations

terms related to specific entities or proper nouns, particularly related to a specific game development company and game titles

oai_token-act-pair · gpt-3.5-turbo

references to the "Arc" series of games and related terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Ö¼

-0.95

ãĤ¦ãĤ¹

-0.78

natal

-0.78

nesday

-0.77

milo

-0.76

matter

-0.75

Todd

-0.75

tons

-0.73

tk

-0.71

ighter

-0.71

POSITIVE LOGITS

Arc

1.22

Arc

1.02

adia

0.93

adian

0.89

 Seraph

0.88

asio

0.76

Corp

0.75

agraph

0.75

 Warden

0.71

oma

0.68

Activations Density 0.009%

terms related to specific entities or proper nouns, particularly related to a specific game development company and game titles

references to the "Arc" series of games and related terms

No Comments

No Known Activations