INDEX

Explanations

explicit mentions or discussions of delegates and their credentials, often related to political conventions

oai_token-act-pair · gpt-3.5-turbo

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

âĹ¼

-0.86

idepress

-0.82

 Mood

-0.77

 Smile

-0.77

 antidepressants

-0.76

 Psychiatric

-0.74

 antidepressant

-0.74

 Mehran

-0.73

 behavioural

-0.72

obyl

-0.72

POSITIVE LOGITS

legates

1.55

 delegates

1.54

 caucuses

1.41

RNC

1.39

 nominating

1.37

 convention

1.32

 delegate

1.30

 caucus

1.23

 ballots

1.21

aucuses

1.20

Activations Density 0.169%

explicit mentions or discussions of delegates and their credentials, often related to political conventions

No Comments

No Known Activations