INDEX

Explanations

phrases related to quantity or degree

oai_token-act-pair · gpt-3.5-turbo

references to specific quantities, types, or characteristics related to objects and situations

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

perse

-0.67

 kidding

-0.65

heid

-0.62

 unden

-0.62

 vain

-0.60

imming

-0.59

 Yourself

-0.57

 Debor

-0.56

BuyableInstoreAndOnline

-0.56

Tro

-0.55

POSITIVE LOGITS

 thresholds

0.75

Ore

0.75

olson

0.70

 threshold

0.69

Keefe

0.66

 certain

0.64

odox

0.62

catentry

0.62

ASON

0.62

 undefined

0.61

Activations Density 0.150%

phrases related to quantity or degree

references to specific quantities, types, or characteristics related to objects and situations

No Comments

No Known Activations