INDEX

Explanations

words related to battles and historical events

oai_token-act-pair · gpt-3.5-turbo

references to historical battles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 referrals

-0.83

employment

-0.76

 assistants

-0.76

 redistributed

-0.76

 giveaways

-0.74

 prosec

-0.74

iblings

-0.74

tarians

-0.73

 admins

-0.73

 entrepreneurs

-0.73

POSITIVE LOGITS

 Mount

1.22

 Normandy

1.13

 Gibraltar

1.12

 Auschwitz

1.12

 Hastings

1.12

Mt

1.11

 Constantinople

1.10

 Bunker

1.10

 Waterloo

1.09

 Malta

1.09

Activations Density 0.150%

words related to battles and historical events

references to historical battles

No Comments

No Known Activations