INDEX

Explanations

phrases related to distinct concepts or ideas, such as opinions, problems, or conflicts

oai_token-act-pair · gpt-3.5-turbo

terms related to physical phenomena and interactions

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

 Hond

-0.69

ones

-0.68

rogens

-0.67

 Bundy

-0.65

 Sections

-0.63

 Bezos

-0.62

rogen

-0.62

 Downs

-0.62

ainers

-0.61

 Highlander

-0.60

POSITIVE LOGITS

âĺ

1.13

âĢ

1.11

[/

1.03

</

0.96

ðŁ

0.88

Â¨

0.86

ãĢ

0.82

.</

0.80

mma

0.78

ðŁĳ

0.77

Activations Density 0.661%

phrases related to distinct concepts or ideas, such as opinions, problems, or conflicts

terms related to physical phenomena and interactions

No Comments

No Known Activations