INDEX

Explanations

action words related to consumption and habits

oai_token-act-pair · gpt-3.5-turbo

phrases relating to caution and responsible behavior

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

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Wellington

-0.72

 honoured

-0.71

 hemisphere

-0.61

 apologised

-0.61

 realised

-0.61

 Cohn

-0.60

 recognised

-0.60

 Hercules

-0.59

 Nasa

-0.59

 Jupiter

-0.57

POSITIVE LOGITS

âĢ

1.81

[/

1.64

</

1.61

ãĢ

1.57

Â»

1.54

âľ

1.52

>>>>

1.50

***

1.48

Â«

1.47

âĺ

1.46

Activations Density 1.762%

action words related to consumption and habits

phrases relating to caution and responsible behavior

No Comments

No Known Activations