INDEX

Explanations

phrases related to prestigious recognition or awards

oai_token-act-pair · gpt-3.5-turbo

references to awards and accolades, particularly in entertainment and journalism

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

 proxies

-0.71

 interactions

-0.70

 spont

-0.68

 anus

-0.68

 verbs

-0.67

 juveniles

-0.67

 cooper

-0.66

kin

-0.65

 mirrors

-0.62

 interaction

-0.62

POSITIVE LOGITS

winning

1.64

eligible

1.36

worthy

1.21

caliber

1.21

nom

1.19

winner

1.17

laden

1.14

favorite

1.11

level

1.10

sized

1.10

Activations Density 0.048%

phrases related to prestigious recognition or awards

references to awards and accolades, particularly in entertainment and journalism

No Comments

No Known Activations