INDEX

Explanations

interrupt-related words and phrases

oai_token-act-pair · gpt-3.5-turbo

instances of the word "interrupt" and its variations in different contexts

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

ndra

-0.95

illac

-0.73

emetery

-0.70

hl

-0.70

hare

-0.70

imb

-0.69

venture

-0.68

hash

-0.68

ramid

-0.67

illet

-0.67

POSITIVE LOGITS

 interrupt

0.97

 interrupts

0.94

 interrupted

0.88

poke

0.85

cffffcc

0.77

interrupted

0.76

 interruption

0.75

aneously

0.74

KP

0.72

ions

0.71

Activations Density 0.020%

interrupt-related words and phrases

instances of the word "interrupt" and its variations in different contexts

No Comments

No Known Activations