INDEX

Explanations

phrases indicating prohibition or advisement against specific actions

oai_token-act-pair · gpt-3.5-turbo

negative imperatives related to restrictions or prohibitions

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

 Tags

-0.80

ãĤ¼ãĤ¦ãĤ¹

-0.78

CRIP

-0.70

LOT

-0.70

licks

-0.69

 Films

-0.67

 slams

-0.67

 Inventory

-0.66

 Starts

-0.66

 Thoughts

-0.65

POSITIVE LOGITS

icably

1.40

 necessarily

1.15

icable

1.07

 adequately

0.94

 permitted

0.92

 disqualified

0.91

 obliged

0.85

withstanding

0.85

 allowed

0.83

 prejud

0.83

Activations Density 0.235%

phrases indicating prohibition or advisement against specific actions

negative imperatives related to restrictions or prohibitions

No Comments

No Known Activations