INDEX

Explanations

the word "trilogy."

oai_token-act-pair · gpt-3.5-turbo

references to trilogies and their installments

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

rooms

-0.82

cius

-0.74

irc

-0.71

 ivory

-0.70

 commons

-0.70

groups

-0.67

ople

-0.66

pots

-0.65

IRC

-0.65

tu

-0.65

POSITIVE LOGITS

 trilogy

1.99

 storyline

1.92

quel

1.92

 thriller

1.90

 reboot

1.62

 finale

1.57

 installment

1.57

 sequel

1.54

 storylines

1.48

 teaser

1.42

Activations Density 0.048%

the word "trilogy."

references to trilogies and their installments

No Comments

No Known Activations