INDEX

Explanations

words related to legal entities, specifically courts and legal procedures

oai_token-act-pair · gpt-3.5-turbo

references to judicial systems and court rulings

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

abwe

-0.76

-0.68

ubric

-0.67

ocument

-0.67

 Sexy

-0.67

Hyd

-0.66

ocom

-0.65

 Marketable

-0.65

 Prosper

-0.64

 Longh

-0.64

POSITIVE LOGITS

 ruled

1.24

 upheld

1.21

 rulings

1.18

 ruling

1.17

 sided

1.15

 justices

1.13

 judge

1.12

 verdict

1.11

 unanimously

1.08

 acquitted

1.05

Activations Density 0.061%

words related to legal entities, specifically courts and legal procedures

references to judicial systems and court rulings

No Comments

No Known Activations