INDEX

Explanations

phrases related to negative or harmful actions or characteristics

oai_token-act-pair · gpt-3.5-turbo

terms related to negative or harmful actions and sentiments

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

inet

-0.94

ais

-0.85

inances

-0.81

rozen

-0.80

ered

-0.79

orius

-0.79

liner

-0.78

alist

-0.78

lique

-0.77

arb

-0.75

POSITIVE LOGITS

 nasty

0.87

 smear

0.82

terday

0.79

 surprises

0.75

 poisons

0.73

 poison

0.72

soever

0.70

mud

0.69

 hello

0.69

 slander

0.66

Activations Density 0.032%

phrases related to negative or harmful actions or characteristics

terms related to negative or harmful actions and sentiments

No Comments

No Known Activations