INDEX

Explanations

phrases indicating strong emotional reactions, especially with negative connotations

oai_token-act-pair · gpt-3.5-turbo

occurrences of ellipses 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

Newsletter

-0.80

 ÂŃ

-0.78

\(\

-0.63

 antip

-0.59

 fluorescent

-0.53

 staffers

-0.52

 living

-0.49

 hypothetical

-0.49

icio

-0.49

 Gleaming

-0.49

POSITIVE LOGITS

..

3.38

..

2.55

......

2.31

....

2.23

.......

2.22

.........

2.18

.....

2.17

........

2.01

âĢ¦..

1.97

...

1.90

Activations Density 0.008%

phrases indicating strong emotional reactions, especially with negative connotations

occurrences of ellipses in the text

No Comments

No Known Activations