INDEX

Explanations

terms related to foreign affairs, intelligence, and surveillance

oai_token-act-pair · gpt-3.5-turbo

references to government and intelligence operations

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

TPS

-0.70

erd

-0.67

agons

-0.65

COMPLE

-0.64

CHAT

-0.64

prototype

-0.63

Cooldown

-0.63

abus

-0.61

proven

-0.60

 intersections

-0.60

POSITIVE LOGITS

 Affairs

0.78

uador

0.73

 Lavrov

0.71

bia

0.70

 espionage

0.69

pora

0.68

 embassies

0.68

 attaches

0.67

 Confederation

0.67

 embassy

0.67

Activations Density 0.078%

terms related to foreign affairs, intelligence, and surveillance

references to government and intelligence operations

No Comments

No Known Activations