INDEX

Explanations

instances where actions are being taken or potential outcomes are being considered

oai_token-act-pair · gpt-3.5-turbo

qualifying adverbs that indicate uncertainty or variability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ayers

-0.82

acas

-0.73

itter

-0.72

achev

-0.71

ierrez

-0.71

kees

-0.70

Ĥİ

-0.69

ente

-0.69

atched

-0.68

iren

-0.68

POSITIVE LOGITS

 outright

0.94

 downright

0.89

 even

0.85

 entire

0.74

 assassinate

0.73

 entirety

0.67

 worse

0.67

 fatally

0.65

 whole

0.63

 vice

0.63

Activations Density 0.160%

instances where actions are being taken or potential outcomes are being considered

qualifying adverbs that indicate uncertainty or variability

No Comments

No Known Activations