INDEX

Explanations

references to documentaries or documentary-related content

oai_token-act-pair · gpt-3.5-turbo

mentions of documentaries

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

cffff

-1.00

hare

-0.80

sembly

-0.75

hap

-0.70

unin

-0.69

cream

-0.68

laws

-0.68

Commerce

-0.67

nyder

-0.67

efficients

-0.67

POSITIVE LOGITS

 filmmaker

1.27

 filmmakers

1.14

 film

1.09

 documentary

1.07

 filmmaking

1.05

film

0.97

 films

0.95

 clip

0.90

 documentaries

0.90

 aired

0.89

Activations Density 0.059%

references to documentaries or documentary-related content

mentions of documentaries

No Comments

No Known Activations