INDEX

Explanations

positions or titles held by individuals

oai_token-act-pair · gpt-3.5-turbo

titles and roles related to leadership positions and professions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 afterward

-0.69

ãĤ©

-0.63

 accordingly

-0.62

 afterwards

-0.61

udos

-0.61

oaded

-0.60

 introdu

-0.59

 drawbacks

-0.58

 wiser

-0.58

 beforehand

-0.57

POSITIVE LOGITS

 Rohingya

0.99

 Palestinian

0.92

zbollah

0.89

 Houth

0.88

 Donetsk

0.87

Syrian

0.86

 Syrian

0.84

 Islamic

0.83

PKK

0.82

 Hamas

0.81

Activations Density 0.910%

positions or titles held by individuals

titles and roles related to leadership positions and professions

No Comments

No Known Activations