INDEX

Explanations

mentions of different headquarters or central offices of organizations

oai_token-act-pair · gpt-3.5-turbo

mentions of various headquarters and their locations

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

 Thrust

-0.67

uel

-0.65

acial

-0.64

icular

-0.63

erer

-0.63

flix

-0.62

assic

-0.62

rieg

-0.62

oven

-0.61

ricular

-0.61

POSITIVE LOGITS

 premises

1.13

 located

1.11

 buildings

1.09

 headquarters

1.06

 building

1.03

 Building

0.99

 offices

0.94

 situated

0.93

 annex

0.91

 Buildings

0.90

Activations Density 0.148%

mentions of different headquarters or central offices of organizations

mentions of various headquarters and their locations

No Comments

No Known Activations