INDEX

Explanations

words related to the thriller genre, often with high activation values

oai_token-act-pair · gpt-3.5-turbo

references to thriller genres in films and literature

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gra

-0.77

anton

-0.77

ral

-0.76

gemony

-0.73

pled

-0.71

ported

-0.70

regulated

-0.70

dated

-0.70

oras

-0.69

cess

-0.69

POSITIVE LOGITS

 thriller

1.44

 Hots

0.81

 Hitman

0.77

 Outbreak

0.77

 unfold

0.77

 Flavoring

0.76

 Hitchcock

0.76

 unfolds

0.76

 screenplay

0.73

 Interstellar

0.73

Activations Density 0.012%

words related to the thriller genre, often with high activation values

references to thriller genres in films and literature

No Comments

No Known Activations