INDEX

Explanations

terms related to conquest or conquering

oai_token-act-pair · gpt-3.5-turbo

references to conquest and domination

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

IFrame

Link

Negative Logits

 Moody

-0.84

email

-0.76

RF

-0.75

ylene

-0.74

>>>

-0.74

 partName

-0.73

HS

-0.71

 Grassley

-0.71

>>

-0.70

IAL

-0.70

POSITIVE LOGITS

 conquer

3.31

 conquered

3.11

 conquering

2.88

 conquest

2.53

 conqu

2.51

 Conquer

2.32

 vanquished

2.14

 Conquest

1.61

 subjug

1.50

 invaders

1.50

Activations Density 0.031%

terms related to conquest or conquering

references to conquest and domination

No Known Activations