INDEX

Explanations

phrases related to legal and governmental actions

oai_token-act-pair · gpt-3.5-turbo

references to legal entities and government-related activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

lehem

-0.80

MON

-0.79

ubes

-0.78

ãĤ©

-0.78

LX

-0.75

RW

-0.75

 Porn

-0.75

onday

-0.74

 Kidd

-0.74

 Monkey

-0.73

POSITIVE LOGITS

Ag

2.33

Ag

2.19

AG

1.80

ag

1.53

 Agents

1.38

Agent

1.36

AG

1.35

 Agency

1.33

agent

1.22

 Agent

1.20

Activations Density 0.205%

phrases related to legal and governmental actions

references to legal entities and government-related activities

No Comments

No Known Activations