INDEX

Explanations

references to incidents or reports of someone being found deceased

oai_token-act-pair · gpt-3.5-turbo

instances of the word "found" along with references to death or discovery in a narrative context

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

adan

-0.81

idth

-0.76

dayName

-0.70

gemony

-0.69

hawks

-0.65

 specificity

-0.65

ichever

-0.64

dos

-0.64

creation

-0.64

osate

-0.64

POSITIVE LOGITS

 dead

1.07

 hanged

0.99

 slumped

0.95

 hanging

0.95

 wandering

0.95

 alive

0.95

 floating

0.95

 abandoned

0.93

 chained

0.92

dead

0.92

Activations Density 0.058%

references to incidents or reports of someone being found deceased

instances of the word "found" along with references to death or discovery in a narrative context

No Comments

No Known Activations