INDEX

Explanations

words related to physical objects, specifically plates

oai_token-act-pair · gpt-3.5-turbo

references to license plates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Ì¶

-0.72

edy

-0.67

nell

-0.66

 Finn

-0.63

 Survivors

-0.62

 Advis

-0.62

inki

-0.62

erness

-0.61

=-=-=-=-

-0.61

 Colleges

-0.60

POSITIVE LOGITS

 plate

4.13

 plates

3.20

plate

3.06

 Plate

2.80

plates

2.25

 plaque

1.29

 mound

1.21

 dish

1.17

 mantle

1.06

 tray

1.00

Activations Density 0.020%

words related to physical objects, specifically plates

references to license plates

No Comments

No Known Activations