INDEX

Explanations

mentions of military or army-related terms

oai_token-act-pair · gpt-3.5-turbo

references to military forces or groups

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

uran

-0.74

ostic

-0.71

eway

-0.70

ricular

-0.70

rics

-0.69

lus

-0.68

ÏĤ

-0.68

âĸ¬

-0.68

lations

-0.67

lic

-0.66

POSITIVE LOGITS

 colonel

0.97

 fatig

0.92

 corps

0.89

 garrison

0.88

 Corps

0.88

 barracks

0.86

 generals

0.86

naires

0.81

 stationed

0.80

chair

0.78

Activations Density 0.024%

mentions of military or army-related terms

references to military forces or groups

No Comments

No Known Activations