INDEX

Explanations

mentions of military ranks, such as 'general'

oai_token-act-pair · gpt-3.5-turbo

references to high-ranking officials, particularly those with the title "general."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 saved

-0.74

 dotted

-0.69

Pak

-0.68

Â®

-0.65

 owned

-0.61

 Room

-0.60

Source

-0.59

 Â®

-0.59

 save

-0.59

 Ranch

-0.58

POSITIVE LOGITS

general

4.18

 general

2.03

 GENERAL

1.78

General

1.76

 General

1.31

 generalized

1.24

 generals

1.04

special

1.02

broad

0.99

personal

0.97

Activations Density 0.011%

mentions of military ranks, such as 'general'

references to high-ranking officials, particularly those with the title "general."

No Comments

No Known Activations