INDEX

Explanations

phrases related to discussing or dealing with various issues

oai_token-act-pair · gpt-3.5-turbo

mentions of addressing issues or topics in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Drift

-0.79

ãĥĪ

-0.70

netflix

-0.65

}}}

-0.63

omer

-0.62

 streak

-0.62

pire

-0.60

 Fruit

-0.60

 Illusion

-0.60

pir

-0.60

POSITIVE LOGITS

 addressing

3.43

 tackling

1.87

 addressed

1.76

 addresses

1.73

 address

1.73

 resolving

1.64

 correcting

1.62

 discussing

1.59

 responding

1.59

 acknowledging

1.58

Activations Density 0.011%

phrases related to discussing or dealing with various issues

mentions of addressing issues or topics in various contexts

No Comments

No Known Activations