INDEX

Explanations

references to specific locations related to military operations, particularly focusing on Syria

oai_token-act-pair · gpt-3.5-turbo

mentions of Syria and related geopolitical contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Chance

-0.71

Lear

-0.66

 Hawaiian

-0.66

 glim

-0.66

 Icelandic

-0.65

 Tennis

-0.65

 philos

-0.64

Companies

-0.64

 Jackets

-0.64

 newcom

-0.64

POSITIVE LOGITS

 ravaged

0.94

 bordering

0.88

 illegally

0.86

 unilaterally

0.83

 region

0.80

 Basin

0.80

 airspace

0.79

 seizing

0.79

 annexed

0.79

 periphery

0.78

Activations Density 0.144%

references to specific locations related to military operations, particularly focusing on Syria

mentions of Syria and related geopolitical contexts

No Comments

No Known Activations