INDEX

Explanations

the word "scan"

oai_token-act-pair · gpt-3.5-turbo

terms related to scanning processes or devices

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 impunity

-0.72

soDeliveryDate

-0.72

ofi

-0.65

Aid

-0.63

doms

-0.60

 perjury

-0.60

 martyr

-0.58

 animosity

-0.58

 Cind

-0.58

inguishable

-0.57

POSITIVE LOGITS

 scans

1.25

 scan

1.16

 scanning

1.04

lator

1.00

 scanned

0.94

lations

0.94

 Scan

0.93

 scanners

0.90

 scanner

0.86

alys

0.83

Activations Density 0.010%

the word "scan"

terms related to scanning processes or devices

No Comments

No Known Activations