INDEX

Explanations

instances of the word "and" followed by other words, particularly when multiple occurrences are close together with high activation values

oai_token-act-pair · gpt-3.5-turbo

connections and conjunctions in sentences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

advertisement

-0.74

arro

-0.69

edia

-0.67

tesy

-0.67

 prosecut

-0.66

ONSORED

-0.63

coat

-0.63

hower

-0.63

theless

-0.62

 bluff

-0.62

POSITIVE LOGITS

 valleys

0.97

uries

0.95

 oranges

0.93

 Soviets

0.80

 necks

0.77

 territories

0.75

 Territories

0.75

 Earthqu

0.73

 Titans

0.72

izons

0.72

Activations Density 0.271%

instances of the word "and" followed by other words, particularly when multiple occurrences are close together with high activation values

connections and conjunctions in sentences

No Comments

No Known Activations