INDEX

Explanations

umpiring-related terms or actions

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

various forms of the word "umpire."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

ings

-0.63

ash

-0.59

 Moonlight

-0.58

 Naomi

-0.57

agher

-0.57

ingly

-0.57

 anonymity

-0.57

ente

-0.55

nings

-0.55

arie

-0.55

POSITIVE LOGITS

sembly

0.68

aucus

0.65

orer

0.65

ilial

0.64

iatrics

0.64

 ACTIONS

0.64

iation

0.62

querque

0.59

FTWARE

0.58

ocene

0.58

Activations Density 0.285%

umpiring-related terms or actions

various forms of the word "umpire."

No Comments

No Known Activations