INDEX

Explanations

references to legal matters or regulations

oai_token-act-pair · gpt-3.5-turbo

phrases related to legal status or legality

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

 Seasons

-0.84

mble

-0.72

pread

-0.70

 Tycoon

-0.69

 Tate

-0.68

ritch

-0.68

 Clip

-0.67

 Wing

-0.67

 Roses

-0.66

 Velocity

-0.66

POSITIVE LOGITS

 enforce

0.91

 obliged

0.83

 sanctioned

0.81

 obligated

0.80

 mandated

0.78

 speaking

0.76

 compliant

0.76

 wedd

0.74

 exting

0.74

inition

0.74

Activations Density 0.026%

references to legal matters or regulations

phrases related to legal status or legality

No Comments

No Known Activations