INDEX

Explanations

proper nouns and named entities

oai_token-act-pair · gpt-3.5-turbo

instances of the symbol '…' indicating continuation or ellipsis in the text

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

``

-0.78

 Edinburgh

-0.62

 Replay

-0.59

recated

-0.57

 Kepler

-0.56

ode

-0.56

 encyclopedia

-0.56

 JSON

-0.55

 Courier

-0.54

New

-0.54

POSITIVE LOGITS

âĢ¦.

3.31

âĢ¦..

3.02

âĢ¦."

2.75

âĢ¦âĢ¦

2.38

âĢ¦âĢ¦âĢ¦âĢ¦

2.29

âĢ¦]

1.99

âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦âĢ¦

1.90

âĢ¦

1.87

âĢ¦"

1.85

—-

1.83

Activations Density 0.017%

proper nouns and named entities

instances of the symbol '…' indicating continuation or ellipsis in the text

No Comments

No Known Activations