INDEX

Explanations

names and ages of individuals, particularly focused on locations

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

âĶĢ

-0.67

cients

-0.65

{\

-0.64

caps

-0.63

 snowball

-0.62

 dictate

-0.61

entimes

-0.61

 lower

-0.60

{}

-0.59

POSITIVE LOGITS

 Jennifer

1.35

 Cheryl

1.35

 Danielle

1.34

 Jason

1.33

 Kenneth

1.32

 Kathleen

1.32

 Brian

1.32

 Kathryn

1.32

 Timothy

1.32

 Jonathan

1.32

Activations Density 0.425%

names and ages of individuals, particularly focused on locations

No Comments

No Known Activations