INDEX

Explanations

details related to specific events, actions, crimes, or statements

oai_token-act-pair · gpt-3.5-turbo

phrases related to criminal incidents and legal contexts

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

liga

-0.86

indust

-0.76

format

-0.76

nergy

-0.75

izons

-0.74

 Patreon

-0.73

 Unicode

-0.72

optim

-0.72

oldown

-0.72

lishes

-0.71

POSITIVE LOGITS

 robbery

1.20

 abduction

1.14

 crime

1.13

 intruder

1.12

 burglary

1.09

 murder

1.08

 burgl

1.06

 altercation

1.04

 Defendant

1.03

 incrim

1.02

Activations Density 0.772%

details related to specific events, actions, crimes, or statements

phrases related to criminal incidents and legal contexts

No Comments

No Known Activations